Comparison of machine learning classifiers for identifying key performance indicators in the Chinese Basketball Association

Comparison of machine learning classifiers for identifying key performance indicators in the Chinese Basketball Association | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Comparison of machine learning classifiers for identifying key performance indicators in the Chinese Basketball Association Rui Dong, Miguel-Angel Gomez-Ruano, Shaoliang Zhang, Xiao Xu, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6230128/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 10 Mar, 2026 Read the published version in BMC Sports Science, Medicine and Rehabilitation → Version 1 posted 10 You are reading this latest preprint version Abstract This study investigated the technical and tactical variables related to success in professional basketball, utilizing data from 3,249 Chinese Basketball Association (CBA) games during the 2018–2022 seasons. Key performance indicators (KPIs) were identified through lasso and stepwise regression, highlighting assists, fouls drawn, offensive rebound percentage, steals, blocks, defensive rebounds, and successful screens-off and pick-and-roll handler’s shots as positively correlated with winning outcomes, while turnovers, missed shots, and fouls demonstrated negative associations. To predict game outcomes, ten machine learning (ML) models were developed, with Artificial Neural Networks (AUC = 0.914), Support Vector Machines (AUC = 0.913), Logistic Regression (AUC = 0.899), and XGBoost (AUC = 0.897) achieving the higher performance. The SHapley Additive exPlanations (SHAP) algorithm further enhanced interpretability, quantifying defensive rebounds, offensive rebound percentage, and missed two-point field attempts consistently as the most influential KPIs. This study provides a comprehensive analysis of game-related and play-type statistics, offering further insights into how they work together for optimizing the multifaceted dynamics of basketball performance. The findings prioritize possession retention through rebounding and effective two-point shot selection in predicting the desired game outcome, presenting how ML-driven KPI identification inform coaching game strategies and refine opponent-specific preparations. basketball key performance indicators CBA play-type machine learning classifiers Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 1 Introduction Systematic evaluation of multidimensional performance indicators constitutes a fundamental requirement in professional basketball analytics [ 1 , 2 ]. Through rigorous analysis of these datasets, coaching staff and performance analysts decode tactical patterns and game dynamics to gain strategic advantages [ 3 , 4 ]. This scientific approach has driven significant methodological advancements in key performance indicators (KPIs) identification research, particularly by developing evidence-based frameworks for tactical optimization and performance enhancement [ 5 , 6 ]. Traditional statistical methods including analysis of variance (ANOVA), discriminant analysis, and principal components analysis (PCA), have formed the methodological foundation for KPI identification in basketball research [ 7 – 10 ]. In the Spanish ACB league, PCA decomposition of game statistics revealed five latent dimensions: scoring efficiency (free-throws, two-point field goals, three-point field-goals), possession control (assists, turnovers), and two rebound-related metrics [ 11 ]. Cluster analysis, combined with hypothesis testing, established three canonical KPIs across European competitions: defensive rebounds, assists, and scoring metrics [ 12 , 13 ]. This pattern emerged consistently in both domestic leagues [ 13 ] and international tournaments [ 14 , 15 ]. Discriminant analysis in NBA studies has identified five key factors distinguishing winners from losers: field goal percentage (SC = 0.71), defensive rebounds (SC = 0.60), offensive rebounds (SC = 0.52), turnovers (SC = 0.45), and steals (SC = 0.42) [ 7 ]. However, these static parameters provide limited explanatory power for the dynamic nature of game processes, primarily offering a general overview of outcome rather than explaining how they evolve [ 4 ]. For example, a high shooting percentage may indicate a player’s shooting ability or effective shot creation by teammates, yet it fails to account for the cooperation and opposition behaviors preceding the shot action. Contemporary basketball research requires a dual-axis evaluation framework that integrates traditional game-related statistics with task-oriented performance indicators [ 16 ]. Within this paradigm, play-type statistics have emerged as a transformative analytical approach, particularly due to their ability to describe pre-shot action executed through individual, paired, and team interactions [ 17 ]. Recent studies have primarily focused on single or a limited set of specific actions-such as transitions [ 18 ], cuts [ 19 ], isolations [ 4 ], and pick-and-roll plays [ 20 , 21 ] across various levels of competitions, analyzing their frequencies and effectiveness within possessions. Findings consistently suggest that the increasing prevalence of pick-and-roll and transition team plays remains a defining trend, with high motion speed and cooperative actions proving generally more effective than individual ball-handler’s maneuvers [ 4 , 22 ]. Marmarinos et al.'s [ 17 ] analysis of 12,376 pick-and-roll sequences in Euroleague basketball demonstrated optimal efficiency under two conditions: (a) when a shot was attempted following two passes from the pick-and-roll action, and (b) when the screener successfully finished at the basket after rolling. Subsequent analyses of the 2020–2021 men’s Euroleague season revealed that the most frequently used possessions and highest point-producing actions were catch-and-shoot plays and pick-and-roll ball handler possessions [ 1 ]. These findings highlight the importance of spatial dynamic, well-coordinated team movements, and highly skilled players in enhancing play execution success in professional basketball. The coach’s decision to elect a particular play type in critical game moments, such as endgame situations, can be a decisive factor in a team’s win or loss [ 23 ]. However, limited studies have integrated play-type statistics to predict overall team success [ 24 ]. Conte et al. [ 3 ] found that NCAA teams performing more hand-offs, but fewer off-ball screens were more likely to outscore their opponents. Similarly, Bustamante et al. [ 24 ] identified higher successful rates of catch-and-shoot plays, cuts, and pick-and-roll (both handler and roller actions) among winning teams in the NBA 2019–2020 season. While these metrics effectively capture certain aspects of performance, they may not fully encapsulate the complexities inherent in professional basketball gameplay when analyzed solely through traditional linear techniques [ 25 ]. Machine learning (ML) algorithms have emerged as powerful tools for predictive modeling in basketball analytics, enabling multivariate analysis of performance indicators, particularly in areas such as training regimens, injury patterns, play-style strategies, and short-, mid-, and long-term match performance evaluation [ 26 – 28 ]. ML algorithms including decision trees (DT), support vector machines (SVM), random forest (RF), artificial neural network (ANN), and XGBoost, have demonstrated superior capability in modeling non-linear interactions within multivariate performance datasets [ 29 – 32 ]. For instance, Leicht et al. [ 25 ] implemented a hybrid analytical framework integrating conditional inference classification trees with logistic regression to identify KPIs associated with Olympic basketball success. A multimodal analysis revealed field-goal percentage, defensive rebounds, steals and turnovers were statistically significant predictors across both linear and non-linear models. In an NBA case study, ensemble modeling combining DT and RF algorithms identified shooting efficiency as the primary KPI, followed by defensive rebounds and opponent turnovers for the Golden State Warriors [ 33 ]. While ML algorithms have demonstrated predictive superiority, model interpretability limitations persist, particularly in quantifying individual indicator contributions to outcome predictions [ 34 ]. To address this, Ouyang et al. [ 35 ] developed a SHAP-XGBoost framework integrating seven ML algorithms (KNN, LightGBM, SVM, RF, Logistic Regression, XGBoost and DT) with SHapley Additive exPlanations (SHAP). The findings identified that field goal percentage, defensive rebounds and turnovers as real-time predictors of NBA game outcomes. The Chinese Basketball Association (CBA), Asia’s premier professional league, exhibits distinct characteristics in tactical implementations, team structures, and stylistic approaches, differentiating it from other leagues [ 36 , 37 ]. Using basic game-related statistics, Cai et al. [ 31 ] developed a hybrid ensemble learning framework based on an SVM model to predict game outcomes during the 2016–2017 regular season. Unlike the NBA and Euroleague, where spacing and three-point shooting have become increasingly dominant over the past decade [ 38 , 39 ], the CBA remains more reliant on mid-range shooting and paint-based plays, as reflected in its lower three-point attempt rate per game [ 36 , 40 ]. Additionally, studies indicate that the CBA employs fewer direct pick-and-roll plays compared to the NBA [ 37 ], which may influence the interpretation of KPIs in this context. These differences necessitate a specialized performance analysis approach, tailored to the distinct gameplay characteristics of the CBA. Previous studies have predominantly relied on traditional linear techniques to explore game-related KPIs, often neglecting comprehensive analyses that integrate both technical and tactical perspectives [ 24 , 41 ]. Furthermore, prior ML-based basketball studies have primarily focused on single or hybrid models, with limited exploration of different ML algorithms to establish a comprehensive and stable set of KPIs across multi-season datasets [ 42 , 43 ]. This limitation affects the objectivity of game outcomes predictions and model interpretability, commonly referred to as “black box” issue, particularly in leagues outside the NBA and Euroleague [ 35 ]. Therefore, the novelty of our study lies in its focus on the CBA league and the application of multipliable ML classifiers for identifying critical KPIs and quantifying their impacts on game outcomes in a more comparative and interpretable manner. Firstly, traditional t-tests and K-means clustering methods are employed to preliminary selected game-related and play-type statistics associated with winning outcomes from the 2018–2022 regular seasons. Subsequently, these parameters undergo essential pre-processing and are merged for feature selection using methods such as lasso and stepwise regression to confirm the final KPIs. Furthermore, ten different types of ML classifiers-including DT, RF, SVM, ANN, GBDT, XGBoost, Adaboost, CatBoost, LightGBM, and LR-are utilized to predict game outcomes based on the identified KPIs. Finally, the SHAP algorithm is implemented to quantify feature importance in the optimal-performing models, providing a more granular interpretation of KPI contributions to game outcomes. 2 Methods 2.1 Sample and variables The data for this study were obtained from Hudl InStat Basketball, a commercially accessible provider based in the USA. A total of 3, 249 games from the 2018–2022 CBA regular seasons were analyzed. The reliability and validity of the system have been verified [ 44 ]. To further assess the reliability of game performance data in the CBA, two experienced coaches were called up to analyze a randomly selected sample of ten games. The inter-observer kappa values and the kappa values of observed data and the data from Hudl InStat Basketball were all higher than 0.91. Table 1 presents the grouping of each parameter and the action definitions provided by InStat. Based on previous studies [ 24 ], 56 basketball game performance parameters were selected and were categorized into two groups: 34 game-related and 22 play-type statistics. Table 1 Selected parameters and definitions Parameters Definitions Game-related statistics Offensive parameters Two-point field goals : field goals shots inside the three-point line or when the player’s foot is touching it Three-point field goals : field goals attempted beyond the three-point line Free throws : a shot from the free-throw line awarded after a foul on the shooter by the opposing team or after exceeding the foul limit Assists : a pass to a teammate that directly leads to a made field goal Turnovers : an action that results in the offensive team losing possession of the ball to the opposing team Fouls drawn : a foul committed by the opponents, stopping the game and followed by either a free throw Offensive rebounds : gaining control of the live ball by the offensive team after their missed shot attempt or free throw Offensive rebounds% : percentage of all available offensive rebounds a team can get. Formula = offensive rebounds / (offensive rebounds + opponent’s defensive rebounds) Defensive parameters Steals : an action where the defending player causes a turnover by taking away the ball Blocks : a defensive action where a player legally deflects a field goal attempt by an offensive player Fouls : a breach of the rules, which can be categorized as personal, technical, unsportsmanlike, flagrant and team fouls Defensive rebounds : gaining control of the live ball by the defensive team after a missed shot attempt or free throw of the opponent Defensive rebounds% : percentage of all available defensive rebounds a team can get. Formula = defensive rebounds / (defensive rebounds + opponent’s offensive rebounds) Play-type statistics Transitions : ball possessions that start at team’s backcourt and last between 4 and 8 seconds Catch and shoot : a play that ends with a jump shot taken at least 3 meters from the rim by a player who controlled the ball less than 2 seconds or did not dribble Catch and drive : a play that ends with a shot taken from 3 meters or less from the rim by a player who dribbled the ball after receiving a pass Drive with shot : a play that ends with a shot taken from 3 meters or less from the rim by a player who dribbled Post up : an offensive player receives the ball within 4.5 meters of the basket with his back to it and attempts a shot Isolations : a play in which a team gives the ball handler space to play one-on-one against his opponent Cuts : a play in which a player attempting to shoot receives the ball while running toward the rim. This includes screens, fast breaks, and situations where a player gets open near the rim Screens-off : a player without the ball legally uses their body to block the defender’s movement, allowing a teammate to break free from the defense On-ball screens : a player sets a screen (pick) for a teammate handling the ball and then slips into the rim (rolls). There are three types: Pick and roll handler - when the ball handler makes a shot attempt; Pick and roll roller - when the screener makes a shot; pick and pop - when the screener stays away from the basket or remains stationary to create a shooting opportunity Contested shot : there is opponent between the rim and a shooter Uncontested shot : there is no opponent between the rim and a shooter 2.2 Statistical analysis Step 1 : A k-means cluster analysis was applied to divide all the games into two groups based on the final score differences: balanced and unbalanced games. Following the inclusion criteria for balanced games, 952 unbalanced games defined as those with final score differences of 19 points or more (considered outliers in this context), were excluded from the analysis. Consequently, the final dataset included 674 games from the 2018–2019 season (246 games were missing due to technical issues with the data provider), 920 games from the 2019–2020 season, 1, 008 games from the 2020–2021 season, and 760 games from the 2021–2022 season. Step 2 : Descriptive statistics (mean and standard deviation) were calculated for each game performance parameter in relation to game outcome (win/loss) for balanced games. An independent samples t-test was conducted to examine differences in each statistic between winning and losing teams. A Pearson’s correlation matrix was then constructed to evaluate collinearity among preliminary performance parameters. Correlation was categorized as follows: negligible or small (|r|<0.1), weak (0.1<|r|<0.3), moderate (0.3<|r|<0.5), strong (0.5<|r|<0.7), very strong (0.7<|r|0.9). Step 3 Lasso and stepwise regression were employed to identify the optimal set of KPIs from a large pool of candidate parameters. These technical and tactical parameters were treated as independent variables, with game outcome coded as the response variable (win = 1, loss = 0). Lasso regression incorporates a regularization term (L1-norm) that shrinks some regression coefficients to zero, retaining only significant predictors. This approach effectively eliminated irrelevant variables and selected significant ones [ 45 ]. The lasso regression process utilized 10-fold cross-validation across 100 “lambda” values, calculating the mean square error (MSE) at each value. The model with the smallest MSE was identified, and a model within one standard error of the minimum MSE was selected [ 46 ]. Stepwise regression was subsequently applied to compare its feature selection results with those of lasso regression, focusing on the overlap of selected features. This method refined the model by incrementally adding statistically significant independent variables and automatically removing insignificant ones [ 47 ]. Step 4 Following the identification of influential KPIs, a variety of mainstream ML classifiers were applied, including DT, RF, SVM, ANN, GBDT, XGBoost, Adaboost, CatBoost, LightGBM, and LR. A stochastic validation strategy was adopted by splitting the data into an 80% training set and 20% testing set. A grid search method was employed for hyperparameter optimization, with the optimal parameter combination subsequently used to construct the ML models. ‌A nested 10-fold cross-validation approach was adopted to prevent overfitting.‌ The performance of these models was evaluated using metrics such as accuracy, recall, precision, F1-score, and the Area Under the Receiver Operating Characteristic Curve (AUC). Finally, the SHAP algorithm was introduced to improve the interpretability of the optimal models’ output. This method computed Shapley values for each feature based on game theory, quantifying their individual contributions to the model’s predictions [48]. All data analyses were conducted using R programming language (version 3.6.3), with the package of “caret” for developing machine learning algorithms. Statistical significance was set at p < 0.05. 3 Results Table 2 and 3 present the results of the descriptive statistics and independent t-tests. Significant differences were found between winning and losing teams for all game-related statistics except for contested and uncontested shot attempts. Furthermore, substantial differences were observed in transitions (made and attempted), catch-and-shoot made, pick-and-pop (PnP) attempted, and eight additional pre-shot combinations between the two groups. Table 2 Descriptive statistics of game-related parameters of winning and losing teams Parameters Win（n=1149） Lose（n=1148） T P two FG made 28.5±5.2 26.2±4.8 -11.199 <0.001 two FG missed 23.7±5.3 25.1±5.2 6.228 <0.001 two FG attempted 52.2±7.6 51.2±7.4 -3.119 <0.001 two FG % 54.7±7.2 51.2±6.7 -12.215 <0.001 three FG made 10.3±3.5 9.5±3.3 -6.144 <0.001 three FG missed 17.4±4.5 19.1±4.8 8.913 <0.001 three FG attempted 27.7±6.5 28.6±6.5 3.184 <0.001 three FG % 37.1±8.7 32.9±8.5 -11.715 <0.001 free throws made 19.8±6.7 17.8±6.2 -7.340 <0.001 free throws missed 6.4±3.1 6.1±3.2 -2.182 0.015 free throws attempted 26.1±8.2 23.9±7.7 -6.867 <0.001 free throws % 75.6±9.9 74.5±10.8 -2.369 0.008 uncontested shot made 6.3±4.0 5.8±3.8 -2.972 0.001 uncontested shot missed 8.9±5.8 9.6±6.2 2.851 0.002 uncontested shot attempted 15.2±9.1 15.4±9.3 0.594 0.276 uncontested shot % 42.3±15.9 38.1±15.2 -6.411 <0.001 contested shot made 29.6±5.6 27.2±5.2 -10.723 <0.001 contested shot missed 32.2±7.5 34.6±7.5 7.646 <0.001 contested shot attempted 61.8±10.5 61.8±10.3 -0.099 0.461 contested shot % 48.2±6.7 44.2±6.2 -14.708 <0.001 assists 22.4±4.9 19.7±4.7 -13.083 <0.001 turnovers 14.3±4.1 15.1±4.3 4.777 <0.001 fouls drawn 24.3±4.6 23.3±4.5 -4.886 <0.001 steals 8.2±3.1 7.6±2.9 -4.799 <0.001 blocks 3.7±2.2 3.1±1.9 -6.283 <0.001 fouls 23.3±4.5 24.3±4.6 4.910 <0.001 offensive rebounds 11.8±4.0 11.2±4.0 -3.579 <0.001 defensive rebounds 30.3±4.7 27.1±4.6 -16.455 <0.001 offensive rebounds % 30.0±8.1 26.7±7.9 -9.953 <0.001 defensive rebounds % 73.3±7.9 70.0±8.1 -9.911 <0.001 Table 3 Descriptive statistics of play-type parameters of winning and losing teams Parameters Win（n=1149） Lose（n=1148） T P transitions made 7.1±3.2 6.2±2.9 -7.161 <0.001 transitions attempted 11.4±4.7 10.9±4.5 -2.302 0.021 catch and shot made 6.5±2.8 5.9±2.8 -5.232 <0.001 catch and shot attempted 17±5.6 16.7±5.7 -1.236 0.217 catch and drive made 2.8±1.8 2.8±1.8 -1.026 0.305 catch and drive attempted 6.7±3.1 7.1±3.2 3.221 0.001 drive with shot made 7.2±3.3 7.1±3.3 -0.708 0.479 drive with shot attempted 14.8±5.9 15.7±6.1 3.379 <0.001 post up made 2.2±2.1 2.2±2.0 -0.887 0.375 post up attempted 4.8±3.6 4.9±3.4 1.318 0.188 isolation made 2.3±2.2 2.2±2.0 -1.252 0.211 isolation attempted 5.4±4.3 5.6±4.2 1.027 0.304 cuts made 4.3±2.4 3.6±2.2 -6.463 <0.001 cuts attempted 6.4±3.2 5.6±3.0 -6.029 <0.001 screens off made 1.9±1.6 1.6±1.5 -3.901 <0.001 screens off attempted 4.1±2.6 4.0±2.7 -0.665 0.506 PnR handlers made 4.8±2.9 4.4±2.6 -2.921 0.004 PnR handlers attempted 11.1±5.3 11±5.1 0.369 0.712 PnR rollers made 1.3±1.4 1.2±1.3 -2.572 0.010 PnR rollers attempted 2.1±1.8 2.0±1.8 -1.369 0.163 PnP made 0.8±1.1 0.9±1.1 0.829 0.407 PnP attempted 2.0±1.9 2.2±2.0 2.531 0.011 Figure 1 illustrates the relationships between the preliminary performance parameters. Strong relationships were observed between free throws made and free throws attempted, offensive rebounds and offensive rebounds%, transitions made and transitions attempted, as well as free throws attempted and fouls drawn. Parameters related to two-point field goals demonstrated very strong correlations. Figure 2a displays the lasso regression path, showing how feature selection was achieved by adjusting the regularization parameter. Figure 2b presents the cross-validation results for lasso regression. By optimizing the regularization parameter λ through ten-fold cross-validation, the model identified the optimal log(λ) value at approximately -4.69 based on the 1-SE rule, selecting a total of 23 parameters. During the process, coefficients for 16 performance parameters such as two FG made and attempted, three FG made and attempted, free throw attempted and missed, contested shot made and missed, uncontested shot made and percentage, offensive rebounds, transitions made, cuts made and attempted, catch-and-drive attempted, PnR rollers made, were shrunk to zero and excluded. Table 3 Identification of key performance indicators based on lasso+logit regression Estimate SE Z P two FG missed -0.21 0.03 -7.16 <0.001 two FG% 0.00 0.02 -0.20 0.844 three FG missed -0.21 0.03 -8.14 <0.001 three FG % 0.06 0.01 4.43 <0.001 free throws made 0.01 0.02 0.50 0.616 free throws% 0.02 0.01 2.45 0.014 assists 0.05 0.02 3.00 0.003 turnovers -0.34 0.02 -15.5 <0.001 fouls drawn 0.08 0.02 3.38 <0.001 uncontested shot missed -0.02 0.01 -1.43 0.153 contested shot% 0.03 0.02 1.42 0.157 steals 0.43 0.03 15.7 <0.001 blocks 0.14 0.03 4.50 <0.001 fouls -0.06 0.01 -3.94 <0.001 defensive rebounds 0.34 0.02 17.5 <0.001 offensive rebound% 0.15 0.01 15.0 <0.001 defensive rebound% 0.01 0.01 1.69 0.091 transitions attempted -0.03 0.02 -1.80 0.071 catch and shot made 0.06 0.03 1.92 0.055 drive with shot -0.01 0.01 -1.21 0.225 screens off made 0.14 0.04 3.15 0.002 PnR handlers made 0.08 0.03 3.12 0.002 PnP attempted 0.07 0.03 2.05 0.040 Table 4 Identification of key performance indicators based on stepwise regression Estimate SE Z P two FG missed -0.231 0.016 -14.236 <0.001 three FG attempted 0.160 0.021 7.642 <0.001 three FG missed -0.457 0.032 -14.292 <0.001 free throws missed -0.091 0.024 -3.885 <0.001 assists 0.060 0.015 4.062 <0.001 turnovers -0.350 0.021 -16.657 <0.001 fouls drawn 0.115 0.017 6.640 <0.001 steals 0.428 0.027 15.730 <0.001 blocks 0.130 0.030 4.357 <0.001 fouls -0.059 0.015 -3.985 <0.001 defensive rebounds 0.354 0.018 19.557 <0.001 offensive rebounds% 0.158 0.010 16.216 <0.001 transitions made 0.101 0.039 2.583 0.010 transitions attempted -0.089 0.027 -3.343 0.001 Screens-off made 0.118 0.041 2.881 0.004 PnR handlers made 0.066 0.023 2.805 0.005 Note: only statistically significant indicators (p < 0.05) are presented in the table Of the 23 parameters retained through lasso regression, defensive rebounds, offensive rebounds%, three FG%, and free throw% emerged as key predictors of CBA game outcomes. Two FG missed, three FG missed, turnovers and fouls, had significant negative impacts, while all other indicators positively influenced game outcomes ( Table 3 ). In comparison, stepwise regression identified 16 significant performance indicators, with transitions attempted showing a significant negative impact ( Table 4 ). A comparative analysis of lasso and stepwise regression revealed a common subset of KPIs selected by both methods ( Figure 3 ). These included two FG missed, three FG missed, assists, turnovers, fouls drawn, and offensive rebounds% (offensive parameters); steals, blocks, fouls and defensive rebounds (defensive parameters) from game-related statistics; as well as screens-off made and PnR handlers made from pre-shot combinations. Table 5 Performance evaluation metrics for the employed ML classifiers Accuracy Precision Recall F1-score AUC DT 65.7% 63.2% 62.6% 0.629 0.703 RF 76.1% 73.9% 75.2% 0.745 0.852 SVM 82.0% 80.8% 80.4% 0.806 0.913 ANN 83.3% 81.9% 82.2% 0.821 0.914 XGBoost 82.4% 81.5% 80.4% 0.809 0.897 GBDT 81.1% 79.3% 80.4% 0.798 0.884 Adaboost 78.9% 78.0% 76.2% 0.771 0.880 CatBoost 81.3% 79.4% 80.8% 0.801 0.895 LightGBM 79.1% 77.8% 77.1% 0.775 0.889 LR 81.3% 79.1% 81.3% 0.802 0.899 Table 5 summarizes the performance metrics of the ten machine learning algorithms employed in this study. Hyperparameter optimization was conducted via grid search, and ten-fold cross-validation was performed to evaluate classifier performance on the test set, ensuring result reliability and robustness. Among the classifiers, ANN achieved the highest accuracy at 83.3%, followed by XGBoost (82.4%) and SVM (82.0%). The DT classifier exhibited the lowest accuracy (65.7%), suggesting its limitations in capturing complex patterns. ANN also achieved the highest F1-score (0.821), reflecting an optimal balance between precision and recall. Furthermore, ANN achieved the highest AUC (0.914), closely followed by SVM with an AUC of 0.913. Overall, these results identified ANN, SVM, XGBoost, and LR as the top-performing classifiers for binary classification task. Figure 4 illustrates the influence of the most informative KPIs on the predictions of the top-performing classifiers. SHAP value quantified the contribution of each feature, revealing that turnovers, two and three FG missed, and fouls were strongly associated with losing probability. Conversely, defensive rebounds, offensive rebounds percentage, assists, fouls drawn, steals, blocks, screens-off made, and PnR handlers made were positively associated with winning probability. Figure 5 compares the feature importance rankings across four ML classifiers (ANN, SVM, XGBoost, and LR). Defensive rebounds (DREB) consistently ranked as the most critical KPI across all classifiers, underscoring their predictive value for game outcomes. Offensive rebounds percentage (OREB%) also ranked among the top three, particularly in ANN and LR. Additionally, missed two FG (2P_Missed) showed higher importance in SVM, XGBoost, and LR, but ranked slightly lower in ANN. Turnovers (TOV) ranked third in ANN but showed reduced importance in other classifiers. Steals (STL), a defensive KPI, ranked in the mid-range for all classifiers, while offensive KPIs like assists (AST) and fouls drawn (FD) appeared in the mid-to-lower range. Defensive indicators such as fouls (FOUL) and blocks (BLK), along with pre-shot combinations like PnR handlers made (PnR_H_Made) and screens-off made（SO_Made）, consistently ranked lower across all classifiers, however, their inclusion highlights their nuanced contribution to predictive frameworks, offering deeper insights into game dynamics. 4 Discussion This study advances basketball performance analytics by systematically identifying KPIs in the CBA through different ML classifiers while contextualizing the findings within the broader context of global basketball leagues. ‌Three primary contributions emerge‌: (1) the identification of CBA-specific KPIs that differ from those in the NBA and European leagues, (2) the integration of tactical parameters (e.g., off- and on-ball screen execution) with traditional game statistics to enhance predictive accuracy, and (3) methodological insights into ML classifier interpretability for quantifying the most predictive indicators of CBA game outcomes and assessing their relative importance. While defensive rebounding and shooting efficiency have been widely recognized as KPIs in professional basketball (NBA, Euroleague), our study reveals that in the CBA, missed two-point attempts have a significantly higher impact on game outcomes than other shooting metrics. This suggests that shot selection dynamics differ in the CBA, with teams placing greater emphasis on mid-range efficiency rather than high-volume three-point shooting. Additionally, our results show that successful pick-and-roll handler plays have a lower predictive value in the CBA than in the NBA, potentially reflecting differences in tactical execution and spacing effectiveness within this league. The prominence of ‌defensive rebounds‌ and offensive rebound percentage‌ as top KPIs aligns with prior studies, reinforcing the universal significance of securing possession after missed shots as a key driver of success [7,49]. The predictive power of defensive rebounding aligns with findings from the NBA, ACB, EuroLeague, and Olympic competitions, underscoring its crucial role in determining professional basketball results [6,12,25,30,50]. Interestingly, while winning teams demonstrated a higher offensive rebounds percentage, they did not necessarily record a greater absolute number of offensive rebounds. This highlights the value of offensive rebounds percentage as a more advanced metric that captures rebounding efficiency rather than just the total count [51]. Similarly, Csátaljay et al. [52] found that offensive rebounds led to significantly higher shooting efficiency and an increased frequency of shooting fouls compared to possession changes, reinforcing the pivotal role of this metric. However, this finding diverges from NBA trends, where offensive rebounds percentage becomes particularly crucial in the later phases of games for winning teams [53]. This suggests that CBA teams may prioritize positioning for high probability putbacks rather than aggressively pursuing every offensive rebound, thereby reducing exposure to transition vulnerabilities. Additionally, the focus of this study on balanced games may account for some discrepancies with from prior research [3,52]. Balanced games typically exhibit different dynamics compared to lopsided contests, potentially altering the relative importance of indicators such as offensive rebounds percentage [24]. Notably, the methodological integration of ‌lasso regression‌, ‌stepwise selection‌, and ‌SHAP-driven ML interpretability‌ addresses a critical gap in basketball analytics. For example, three FG% and free throws% emerged as significant indicators in lasso regression but not in stepwise regression. These discrepancies highlight the importance of employing complementary methods to effectively identify KPIs from a large pool of candidate variables [43]. Traditional studies often rely on single league dataset or linear model, limiting the generalizability of findings [7,24]. By contrast, our multi-classifier framework including ANN, SVM, XGBoost, LR revealed that ‌missed two-point attempts‌-a metric rarely prioritized in NBA analyses-were more predictive of CBA outcomes than successful shots. This finding underscores the ‌CBA’s unique shot selection dynamics‌: teams with fewer missed two-point attempts likely demonstrate superior decision-making in spatially constrained offensive systems, a pattern less pronounced in leagues with higher three-point utilization seen in ACB and NBA leagues [49]. These insights validate the need for league-specific modeling, as KPIs derived from NBA data may misrepresent tactical priorities in Asian basketball contexts. Other game-related statistics, such as blocks, turnovers, and fouls, also exhibited associations with winning likelihood, although these KPIs have not always been consistently emphasized [ 5 , 25 ]. While blocks and fouls ranked relative lower in importance, Sampaio et al. [ 54 ] reported that, in addition to shooting performance, recovered possessions-such as steals, blocks, and fewer turnovers-were among the primary contributors to point differentials during the 2008 Beijing Olympic Games. The ability to recover possession and convert it into effective scoring opportunities was a hallmark of team USA’s dominance in this tournament. The paradoxical role of fouls in CBA outcomes highlights cultural and tactical distinctions [ 36 ]. While losing teams committed more defensive fouls, a pattern consistent with NBA trends [ 5 ], the weaker predictive value of fouls in determining loss likelihood partly suggests that the ‌strategic fouling is less systematically employed‌ in the CBA compared to elite international competitions [55]. Although fouls provide opponents with free-throw opportunities, they may also indicate intense defensive pressure [56]. Strategic fouling can serve as an effective tactic for disrupting the opponent’s rhythm, limiting fast-break opportunities, and neutralizing tactical setups [57]. However, in the CBA, the lower reliance on intentional fouling in late-game scenarios suggests a distinct coaching philosophy that prioritizes maintaining game tempo over disrupting play. This difference may stem from CBA coaching principles favoring a faster-paced offensive transitions, thereby reducing the frequency of “foul-to-stop-clock” strategies [29]. Similarly, the limited predictive power of blocks contrasts with their significance in the NBA, where elite rim protectors exert a disproportionate influence on game outcomes [50]. This divergence may be attributed to the CBA’s emphasis on team-oriented defense rather than individual shot-blocking prowess‌, a characteristic that aligns with broader trends in Asian basketball development philosophies. This study also examined tactical parameters critical to team success alongside game-related statistics. Among these, only ball screen play-types significantly increased the likelihood of winning, underscoring the importance of this commonly used tactical behavior in high-level basketball competitions [ 58 ]. This study further emphasizes the significance of executing more PnR handler’s successful shots for success compared to other on-ball screen types (e.g., roll-in or pop-out), aligning with findings from NBA studies [ 1 ]. However, this result contrasts with Bustamante et al. [ 24 ], who found that PnR handler made plays had negligible predictive value in NBA games. Variations in offensive strategies may explain these differences. For instance, coaches frequently encourage players to attack the basket aggressively after executing a PnR, making the ball handler’s shot a central component of this tactic [ 17 ]. This strategy is particularly evident in teams with elite individual scorers, who often rely on isolation-based finishes to capitalize on one-on-one advantages [ 2 ]. These advantages become particularly valuable in high-pressure situations, such as when structured team play breaks down or the shot clock winds down [ 23 ]. While ball screens are widely recognized as a cornerstone of modern offenses in professional basketball competitions [ 4 , 17 ], their lower predictive value in this study highlights unique tactical challenges in the CBA. Specifically, off-ball screen sequences-which require precise coordination between passers, cutters, and screeners to generate uncontested shots-appear to be either underutilized or inefficiently executed in CBA gameplay [ 59 ]. Unlike in European-style offenses, where high-IQ playmakers and sharpshooters excel in reading defensive rotations and exploiting screens to create open looks for elite perimeter shooters [ 60 ], CBA teams may face challenges in spacing, execution speed, and defensive adaptations, which limit the overall impact of these actions. This is evidenced by the low success rate of off-ball screens, even among winning teams. Furthermore, variations in defensive schemes, such as the frequent use of switching defenses, may further reduce opportunities for executing multi-step off-ball actions, forcing teams to rely more on individual shot creation [ 1 ]. These results also reinforce the notion that it is not merely the quantity of successful screens but rather their finishing efficiency that determines team success [ 24 ]. The similar shooting performance of CBA teams in both contested and uncontested scenarios suggests that screening actions do not consistently generate a significant scoring advantage. This inefficiency diminishes the tactical payoff of screening efforts, rendering them statistically less impactful [ 59 ]. Finally, the limited importance of ‌successful off-ball screens and PnR handler plays ‌in CBA outcomes highlights a critical divergence from global basketball trends. As Marmarinos et al. [17] observed, cooperative actions among players, particularly those involving multiple interconnected plays, are generally more effective than isolated actions by ball handlers. This suggests that CBA teams may benefit from enhancing offensive sequences that focus on greater synergy and continuity, enabling additional scoring opportunities for teammates in optimal shooting positions. 5 Conclusion and future research lines This study aimed to systematically identify KPIs in the CBA using various machine learning classifiers. By integrating traditional game-related statistics with tactical play-type metrics, the study provided data-driven insights into winning strategies in the CBA. The findings highlight the importance of possession retention through rebounding and effective two-point shot selection as key predictors of game outcomes. This study provides a roadmap for integrating ML-driven performance analysis into team decision-making processes. Additionally, automated scouting reports, based on based on identified CBA-specific performance indicators, enable coaching staff to optimize player roles, adjust rotation strategies, and refine opponent-specific preparations. Future studies should expand on these findings by examining ML explainability in live game scenarios, thereby enhancing their practical usability for coaching applications. Specifically, incorporating temporal sequence modeling (e.g., recurrent neural networks) could capture in-game momentum shifts and offer deeper insights into how KPIs evolve over different game phases. Another key direction is the integration of ML models into coaching software. When combined with markerless optical tracking systems, teams can automate possession-based performance analysis and evaluate the impact of off-ball movements and pick-and-roll strategies. Finally, a cross-league comparative analysis incorporating data from the NBA, Euroleague, and other professional leagues, could provide a broader validation of CBA-specific KPIs, contributing to the development of universal framework for performance optimization in professional basketball. Declarations Ethics approval and consent to participate Not applicable. Consent for publication Not applicable. Availability of data and materials All data generated or analysed during this study are included in this published article [and its supplementary information files]. Competing interests The authors have no conflicts of interest to declare that are relevant to the content of this article. Funding The author received no financial support for the research. Authors' contributions RD: Conceptualization, Investigation, Validation, Visualization, Writing – original draft. MG: Methodology, Formal analysis, Writing – review & editing. SZ: Methodology, Formal analysis, Writing – review & editing. XX: Software, Writing – review & editing. ZL: Data curation, Writing – review & editing. BY: Visualization, Writing – review & editing. HL: Conceptualization, Project administration, Resources, Writing – review & editing. Acknowledgements Comments and suggestions from the anonymous reviewers which improved the quality of this paper are appreciated. References Akinci Y. Examining the Differences Between Playoff Teams and Non-Playoff Teams in Men’s Euroleague; Play-Type Statistics Perspective. Sage Open. 2023;13(4):21582440231220155. 10.1177/21582440231220155 . Gómez MÁ, Battaglia O, Lorenzo A, Lorenzo J, Jiménez S, Sampaio J. Effectiveness during ball screens in elite basketball games. J Sports Sci. 2015;33(17):1844–52. 10.1080/02640414.2015.1014829 . Conte D, Tessitore A, Gjullin A, Mackinnon D, Lupo C, Favero T. Investigating the game-related statistics and tactical profile in NCAA division I men’s basketball games. Biol Sport. 2018;35(2):137–43. 10.5114/biolsport.2018.71602 . Christmann J, Akamphuber M, Müllenbach AL, Güllich A. Crunch time in the NBA – The effectiveness of different play types in the endgame of close matches in professional basketball. Int J Sports Sci Coaching. 2018;13(6):1090–9. 10.1177/1747954118772485 . Zhou W, Sansone P, Jia Z, Gomez MA, Li F. Determining the key performance indicators on game outcomes in NBA based on quantile regression analysis. Int J Perform Anal Sport. 2024;0(0):1–16. 10.1080/24748668.2024.2325846 . Çene E. What is the difference between a winning and a losing team: insights from Euroleague basketball. Int J Perf Anal Spor. 2018;18(1):55–68. 10.1080/24748668.2018.1446234 . Cabarkapa D, Deane MA, Fry AC et al. Game statistics that discriminate winning and losing at the NBA level of basketball competition. Abián-Vicén J, ed. PLoS ONE . 2022;17(8): e0273427. 10.1371/journal.pone.0273427 Mikołajec K, Banyś D, Żurowska-Cegielska J, Zawartka M, Gryko K. How to Win the Basketball Euroleague? Game Performance Determining Sports Results During 2003–2016 Matches. J Hum Kinetics. 2021;77:287–96. 10.2478/hukin-2021-0050 . Conte D, Lukonaitienė I. Scoring Strategies Differentiating between Winning and Losing Teams during FIBA EuroBasket Women 2017. Sports . Published online. 2018. 10.3390/sports6020050 Teramoto M, Cross CL, Rieger RH, Maak TG, Willick SE. Predictive Validity of National Basketball Association Draft Combine on Future Performance. J Strength Conditioning Res. 2018;32(2):396–408. 10.1519/JSC.0000000000001798 . Sampaio J, Drinkwater EJ, Leite NM. Effects of season period, team quality, and playing time on basketball players’ game-related statistics. Eur J Sport Sci. 2010;10(2):141–9. 10.1080/17461390903311935 . Ibáñez SJ, García J, Feu S, Lorenzo A, Sampaio J. Effects of consecutive basketball games on the game-related statistics that discriminate winner and losing teams. J Sport Sci Med. 2009;(8):458–62. Gómez MÁ, Lorenzo A, Sampaio J, Ibáñez SJ, Ortega E. Game-Related Statistics that Discriminated Winning and Losing Teams from the Spanish Men’s Professional Basketball Teams. Coll Antropol. 2008;32(2):315–9. Lorenzo A, Gómez MÁ, Ortega E, Ibáñez SJ. Game related statistics which discriminate between winning and losing under-16 male basketball games. J Sport Sci Med. 2010;(9):664–8. Csataljay G, O’Donoghue P, Hughes M, Dancs H. Performance indicators that distinguish winning and losing teams in basketball. Int J Perform Anal Sport. 2009;9(1):60–6. 10.1080/24748668.2009.11868464 . Gómez MA, Ibáñez SJ. The use of classification and regression tree when classifying winning and losing basketball teams. Kinesiology. 2017;49(1):47. 10.26582/k.49.1.9 . Marmarinos C, Apostolidis N, Kostopoulos N, Apostolidis A. Efficacy of the pick and roll offense in top level European basketball teams. J Hum Kinet. 2016;51(1):121–9. 10.1515/hukin-2015-0176 . Conte D, Favero T, Niederhausen M, Capranica L, Tessitore A. Determinants of the effectiveness of fast break actions in elite and sub-elite Italian men’s basketball games. bs . 2017; 2:177–183. 10.5114/biolsport.2017.65337 Zukolo Z, Dizdar D, Selmanović A, Vidranski T. The role of finishing actions in the final result of the basketball match. Sport Sci. 2019;19:90–5. Koutsouridis C, Karamousalidis G, Galazoulas C. The efficacy of High Pick and Roll in relation to the defence’s reaction and its effect on the result of the game. Int J Perform Anal Sport. 2018;18(4):554–67. 10.1080/24748668.2018.1501983 . Lorenzo Calvo J, Menéndez García A, Navandar A. Analysis of mismatch after ball screens in Spanish professional basketball. Int J Perf Anal Spor. 2017;17(4):555–62. 10.1080/24748668.2017.1367999 . Matulaitis K, Bietkis T. Prediction of Offensive Possession Ends in Elite Basketball Teams. Int J Environ Res Public Health. 2021;18(3):1083. 10.3390/ijerph18031083 . Foteinakis P, Pavlidou S, Stavropoulos N. Analysis of the effectiveness of different play types in the end of game possessions of close EuroLeague matches. J Hum Sport Exerc. 2024;19(2):617–30. 10.55860/rj38hm02 . Bustamante-Sánchez A, Gomez-Ruano MA, Clemente-Suárez VJ, Jiménez-Sáiz SL. Pre-shot combinations and game-related statistics discriminating between winners and losers depending on the game location during the NBA COVID-19 season. Front Physiol. 2022;13:949445. 10.3389/fphys.2022.949445 . Leicht AS, Gómez MA, Woods CT. Explaining Match Outcome during The Men’s Basketball Tournament at The Olympic Games. J Sports Sci Med. 2017;(16):468–73. Sarlis V, Chatziilias V, Tjortjis C, Mandalidis D. A Data Science approach analysing the Impact of Injuries on Basketball Player and Team Performance. Inform Syst. 2021;99:101750. 10.1016/j.is.2021.101750 . Sarlis V, Tjortjis C. Sports analytics — Evaluation of basketball players and team performance. Inform Syst. 2020;93:101562. 10.1016/j.is.2020.101562 . Tian C, De Silva V, Caine M, Swanson S. Use of Machine Learning to Automate the Identification of Basketball Strategies Using Whole Team Player Tracking Data. Appl Sci. 2019;10(1):24. 10.3390/app10010024 . Qiu M, Zhang S, Yi Q, Zhou C, Zhang M. The influence of momentum on the game outcome while controlling for game types in basketball. Front Psychol. 2024;15:1412840. 10.3389/fpsyg.2024.1412840 . Thabtah F, Zhang L, Abdelhamid N. NBA Game Result Prediction Using Feature Analysis and Machine Learning. Ann Data Sci. 2019;6(1):103–16. 10.1007/s40745-018-00189-x . Cai W, Yu D, Wu Z, Du X, Zhou T. A hybrid ensemble learning framework for basketball outcomes prediction. Physica A. 2019;528:121461. 10.1016/j.physa.2019.121461 . Harris AR, Roebber PJ. NBA team home advantage: Identifying key factors using an artificial neural network. PLoS ONE. 2019;14(7):e0220630. 10.1371/journal.pone.0220630 . Migliorati M. Detecting drivers of basketball successful games: an exploratory study with machine learning algorithms. Electron J Appl Stat Anal. 2020;13(2):454–73. 10.1285/I20705948V13N2P454 . Ekanayake IU, Meddage DPP, Rathnayake U. A novel approach to explain the black-box nature of machine learning in compressive strength predictions of concrete using Shapley additive explanations (SHAP). Case Stud Constr Mater. 2022;16:e01059. 10.1016/j.cscm. 2022.e01059 . Ouyang Y, Li X, Zhou W et al. Integration of machine learning XGBoost and SHAP models for NBA game outcome prediction and quantitative analysis methodology. Ribeiro HV, ed. PLoS ONE . 2024;19(7): e0307478. 10.1371/journal.pone.0307478 Wang X, Han B, Zhang S, Zhang L, Lorenzo Calvo A, Gomez MÁ. The Differences in the Performance Profiles Between Native and Foreign Players in the Chinese Basketball Association. Front Psychol. 2022;12. 10.3389/fpsyg.2021.788498 . Chen R, Zhang M, Xu X. Modeling the influence of basketball players’ offense roles on team performance. Front Psychol. 2023;14. 10.3389/fpsyg.2023.1256796 . Gómez MÁ, Medina R, Leicht AS, Zhang S, Vaquera A. The Performance Evolution of Match Play Styles in the Spanish Professional Basketball League. Appl Sci. 2020;10(20):7056. 10.3390/app10207056 . Zhang S, Lorenzo A, Woods CT, Leicht AS, Gómez MA. Evolution of game-play characteristics within-season for the National Basketball Association. Int J Sports Sci Coaching. 2019;14(3):355–62. 10.1177/1747954119847171 . Li Mzi, Lu Y, Chen Z, hao, Yang Q, Tao Yliu. Analysis of Key Performance Indicators and Major Winning Rules in Chinese Basketball Professional League. Published online Oct. 2023;10. 10.21203/rs.3.rs-3407093/v1 . Gomez MA, Gasperi L, Lupo C. Performance analysis of game dynamics during the 4th game quarter of NBA close games. Int J Perform Anal Sport. 2016;16(1):249–63. 10.1080/24748668.2016.11868884 . Papageorgiou G, Sarlis V, Tjortjis C. Evaluating the effectiveness of machine learning models for performance forecasting in basketball: a comparative study. Knowl Inf Syst. 2024;66(7):4333–75. 10.1007/s10115-024-02092-9 . Plakias S, Kokkotis C, Pantazis D, Tsatalas T. Comparative analysis of key performance indicators in Euroleague and national basketball leagues. J Phys Educ Sport. 2024;24(6):1360–72. 10.7752/jpes.2024.06154 . Bustamante-Sánchez Á, Gómez MA, Jiménez-Saiz SL. Game location effect in the NBA: A comparative analysis of playing at home, away and in a neutral court during the COVID-19 season. Int J Perf Anal Spor. 2022;22(3):370–81. 10.1080/24748668.2022.2062178 . Krzyszkowski J, Kipp K. Prediction of throwing distance in the men’s and women’s javelin final at the 2017 IAAF world championships. J Sports Sci. 2021;39(9):1055–9. 10.1080/02640414.2020.1856462 . Kolodziej M, Groll A, Nolte K, et al. Predictive modeling of lower extremity injury risk in male elite youth soccer players using least absolute shrinkage and selection operator regression. Scandinavian Med Sci Sports. 2023;33(6):1021–33. 10.1111/sms.14322 . Scanlan A, Humphries B, Tucker PS, Dalbo V. The influence of physical and cognitive factors on reactive agility performance in men basketball players. J Sports Sci. 2014;32(4):367–74. 10.1080/02640414.2013.825730 . Li Z. Extracting spatial effects from machine learning model using local interpretation method: An example of SHAP and XGBoost. Comput Environ Urban Syst. 2022;96:101845. 10.1016/j.compenvurbsys.2022.101845 . Stavropoulos N, Kolias P, Papadopoulou A, et al. Game related predictors discriminating between winning and losing teams in preliminary, second and final round of basketball world cup 2019. Int J Perform Anal Sport. 2021;21(3):383–95. 10.1080/24748668.2021.1901437 . Zhang S, Lorenzo A, Zhou C, Cui Y, Gonçalves B, Angel Gómez M. Performance profiles and opposition interaction during game-play in elite basketball: evidences from National Basketball Association. Int J Perform Anal Sport. 2019;19(1):28–48. 10.1080/24748668.2018.1555738 . Kubatko J, Oliver D, Pelton K, Rosenbaum DT. A Starting Point for Analyzing Basketball Statistics. Journal of Quantitative Analysis in Sports . 2007;3(3): Article 1. 10.2202/1559-0410.1070 Csátaljay G, James N, Hughes M, Dancs H. Analysis of influencing factors behind offensive rebounding performance in elite basketball. Int J Sports Sci Coa. 2017;12(6):774–81. 10.1177/1747954117738900 . Malarranha J, Figueira B, Leite N, et al. Dynamic Modeling of Performance in Basketball. Int J Perform Anal Sport. 2013;13(2):377–87. 10.1080/24748668.2013.11868655 . Sampaio J, Lago C, Drinkwater EJ. Explanations for the United States of America’s dominance in basketball at the Beijing Olympic Games (2008). Journal of Sports Sciences . 2010;28(2):147–152. 10.1080/02640410903380486 Ruano M, Avugos S, Ortega E, Bar-Eli M. Adverse effects of technical fouls in elite basketball performance. Biology Sport. 2019;36:155–60. https://doi.org/10.5114/biolsport.2019.83505 . Sampaio J, Leser R, Baca A, et al. Defensive pressure affects basketball technical actions but not the time-motion variables. J Sport Health Sci. 2016;5:375–80. 10.1016/j.jshs.2015.01.011 . Gómez Mángel, Lorenzo A, Jiménez S, Navarro RM, Sampaio J. Examining Choking in Basketball: Effects of Game Outcome and Situational Variables during Last 5 Minutes and Overtimes. Percept Mot Skills. 2015;120(1):111–24. 10.2466/25.29.PMS.120v11x0 . Gómez MA, Lorenzo A, Ibañez SJ, Sampaio J. Ball possession effectiveness in men’s and women’s elite basketball according to situational variables in different game periods. J Sport Sci. 2013;31(14):1578–87. 10.1080/02640414.2013.792942 . Stavropoulos N, Papadopoulou A, Kolias P. Evaluating the Efficiency of Off-Ball Screens in Elite Basketball Teams via Second-Order Markov Modelling. Mathematics. 2021;9(16):1991. 10.3390/math9161991 . Foteinakis P, Pavlidou S. Positional Differences in the Efficacy of Critical End-of-Game Possessions in EuroLeague Basketball. Sport Mont. 2024;22(2):25–31. 10.26773/smj.240704 . Additional Declarations No competing interests reported. Supplementary Files rawdata.xlsx Cite Share Download PDF Status: Published Journal Publication published 10 Mar, 2026 Read the published version in BMC Sports Science, Medicine and Rehabilitation → Version 1 posted Editorial decision: Revision requested 25 Nov, 2025 Reviews received at journal 24 Nov, 2025 Reviewers agreed at journal 28 Jul, 2025 Reviews received at journal 15 Apr, 2025 Reviewers agreed at journal 24 Mar, 2025 Reviewers agreed at journal 22 Mar, 2025 Reviewers invited by journal 22 Mar, 2025 Editor assigned by journal 21 Mar, 2025 Submission checks completed at journal 20 Mar, 2025 First submitted to journal 20 Mar, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6230128","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":433153301,"identity":"c489119b-c7e0-4235-b5b1-4629ab6767aa","order_by":0,"name":"Rui Dong","email":"","orcid":"","institution":"South China Normal University","correspondingAuthor":false,"prefix":"","firstName":"Rui","middleName":"","lastName":"Dong","suffix":""},{"id":433153304,"identity":"f323ad88-8b7d-4450-8b32-41da986c201d","order_by":1,"name":"Miguel-Angel Gomez-Ruano","email":"","orcid":"","institution":"European University of Madrid","correspondingAuthor":false,"prefix":"","firstName":"Miguel-Angel","middleName":"","lastName":"Gomez-Ruano","suffix":""},{"id":433153307,"identity":"57f510c3-9ca7-47bc-9e90-db5ee414282c","order_by":2,"name":"Shaoliang Zhang","email":"","orcid":"","institution":"Tsinghua University","correspondingAuthor":false,"prefix":"","firstName":"Shaoliang","middleName":"","lastName":"Zhang","suffix":""},{"id":433153313,"identity":"95f6bf57-4ee5-4279-bce6-30ca59251d18","order_by":3,"name":"Xiao Xu","email":"","orcid":"","institution":"Dalian University","correspondingAuthor":false,"prefix":"","firstName":"Xiao","middleName":"","lastName":"Xu","suffix":""},{"id":433153314,"identity":"863116e0-e7ef-4aa7-825d-22a1d487eda8","order_by":4,"name":"Zhongbiao Liu","email":"","orcid":"","institution":"South China Normal University","correspondingAuthor":false,"prefix":"","firstName":"Zhongbiao","middleName":"","lastName":"Liu","suffix":""},{"id":433153315,"identity":"5ccf097e-2c9a-4de8-8a4b-347b1e396dfa","order_by":5,"name":"Bin Yang","email":"","orcid":"","institution":"South China Normal University","correspondingAuthor":false,"prefix":"","firstName":"Bin","middleName":"","lastName":"Yang","suffix":""},{"id":433153316,"identity":"b803f299-4b84-4080-8024-6506e94863c7","order_by":6,"name":"Hongyou Liu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA40lEQVRIiWNgGAWjYHAD5mNgio2deC1saQwMCUCKmXgtPGZgLQyEtMj3nzFg5qk4nNgvkfPtwccf2+T5mBkYP3zMwa3F4ABIy5nDiTNn5G43nJFw27CNmYFZcuY2PFoYewyYedsOJ264kbtNmifhNiNQCxszLx4t8s08MC05z0Ba7AlqYTiG0MIG0pJIUIvBGbaCg3POpBvP7HlmJjkj7XZyGzNjM16/yPcf3vjgTYW1bD978jOJDza3bee3Nx/88BGfw4DgEA8Dg2MDgs/YgEslQskPBgZ7gqpGwSgYBaNg5AIAKVxM185UvC8AAAAASUVORK5CYII=","orcid":"","institution":"South China Normal University","correspondingAuthor":true,"prefix":"","firstName":"Hongyou","middleName":"","lastName":"Liu","suffix":""}],"badges":[],"createdAt":"2025-03-15 03:38:03","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6230128/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6230128/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s13102-026-01620-0","type":"published","date":"2026-03-10T15:58:09+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":79591435,"identity":"2744c026-32e3-4788-a078-601dcc42b779","added_by":"auto","created_at":"2025-03-31 13:12:08","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":122676,"visible":true,"origin":"","legend":"\u003cp\u003ePearson’s correlation matrix of preliminary performance parameters\u003c/p\u003e","description":"","filename":"Picture1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6230128/v1/8cf305c4580ea45e4b81618b.jpg"},{"id":79590301,"identity":"e02784a6-148a-4d8b-b7b5-1462221004b4","added_by":"auto","created_at":"2025-03-31 13:04:08","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":35803,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCross-validation curve of lasso regression. \u003c/strong\u003eIn figure \u003cstrong\u003e2a\u003c/strong\u003e, each line represents the regression coefficient of a feature. As the alpha value increases, unimportant features were gradually excluded, leaving only those significantly influencing game predictions. In figure \u003cstrong\u003e2b\u003c/strong\u003e, the horizontal axis represents the natural logarithm of alpha [log(λ)], the vertical axis reflects the model error, and the top horizontal axis shows the number of indicators selected by the model. A ten-fold cross-validation was used to identify the optimal alpha value corresponding to the minimum mean squared error (blue dashed line) and then apply the 1-SE rule (i.e., choosing the most parsimonious model with an error no more than one standard error above the minimum) to identify a more conservative alpha value (red dashed line).\u003c/p\u003e","description":"","filename":"2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6230128/v1/8a7451b0b5d68cb44ac4d90e.jpg"},{"id":79591436,"identity":"5aa0c12c-cf2b-471d-8b10-1a84fbd2bf91","added_by":"auto","created_at":"2025-03-31 13:12:08","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":91418,"visible":true,"origin":"","legend":"\u003cp\u003eFeature Selection Comparison of Lasso and Stepwise Regression\u003c/p\u003e","description":"","filename":"3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6230128/v1/176352e2eb257101e73d7e9c.jpg"},{"id":79590306,"identity":"74278500-e4ac-4a1e-be43-7872a9fec399","added_by":"auto","created_at":"2025-03-31 13:04:08","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":64811,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSummary of SHAP value of top-performing ML classifiers.\u003c/strong\u003e The color represents the risk level associated with match observations, with red indicating higher values and blue indicating lower values.\u003c/p\u003e","description":"","filename":"4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6230128/v1/adfa4c74ce633a10f31a1dba.jpg"},{"id":79590303,"identity":"cd65466d-6ef3-4c43-82e3-219236b02574","added_by":"auto","created_at":"2025-03-31 13:04:08","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":49553,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFeature ranking comparison across four top-performing ML classifiers.\u003c/strong\u003e Each circle represents a KPI, with its vertical position indicating the importance within the respective model. Color ranges from dark blue (high importance) to light blue (low importance), with the same color representing identical KPIs across models to emphasize ranking sequence. Arrows connect the same KPIs, visually representing each KPI’s importance varies among different classifiers.\u003c/p\u003e","description":"","filename":"5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6230128/v1/ceeb34b39ce79318aecc62cd.jpg"},{"id":104739587,"identity":"84d31b08-79bb-47af-bb03-93fd26601471","added_by":"auto","created_at":"2026-03-16 16:09:51","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1718966,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6230128/v1/1e0c02ce-41d6-48dc-9099-bc6aa9258858.pdf"},{"id":79590310,"identity":"86c66289-f934-483d-a792-87c449269995","added_by":"auto","created_at":"2025-03-31 13:04:08","extension":"xlsx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":1051194,"visible":true,"origin":"","legend":"","description":"","filename":"rawdata.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6230128/v1/6f206cfebab4de6fbf3f3d53.xlsx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Comparison of machine learning classifiers for identifying key performance indicators in the Chinese Basketball Association","fulltext":[{"header":"1 Introduction","content":"\u003cp\u003eSystematic evaluation of multidimensional performance indicators constitutes a fundamental requirement in professional basketball analytics [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. Through rigorous analysis of these datasets, coaching staff and performance analysts decode tactical patterns and game dynamics to gain strategic advantages [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. This scientific approach has driven significant methodological advancements in key performance indicators (KPIs) identification research, particularly by developing evidence-based frameworks for tactical optimization and performance enhancement [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eTraditional statistical methods including analysis of variance (ANOVA), discriminant analysis, and principal components analysis (PCA), have formed the methodological foundation for KPI identification in basketball research [\u003cspan additionalcitationids=\"CR8 CR9\" citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. In the Spanish ACB league, PCA decomposition of game statistics revealed five latent dimensions: scoring efficiency (free-throws, two-point field goals, three-point field-goals), possession control (assists, turnovers), and two rebound-related metrics [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. Cluster analysis, combined with hypothesis testing, established three canonical KPIs across European competitions: defensive rebounds, assists, and scoring metrics [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. This pattern emerged consistently in both domestic leagues [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e] and international tournaments [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. Discriminant analysis in NBA studies has identified five key factors distinguishing winners from losers: field goal percentage (SC\u0026thinsp;=\u0026thinsp;0.71), defensive rebounds (SC\u0026thinsp;=\u0026thinsp;0.60), offensive rebounds (SC\u0026thinsp;=\u0026thinsp;0.52), turnovers (SC\u0026thinsp;=\u0026thinsp;0.45), and steals (SC\u0026thinsp;=\u0026thinsp;0.42) [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. However, these static parameters provide limited explanatory power for the dynamic nature of game processes, primarily offering a general overview of outcome rather than explaining how they evolve [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. For example, a high shooting percentage may indicate a player\u0026rsquo;s shooting ability or effective shot creation by teammates, yet it fails to account for the cooperation and opposition behaviors preceding the shot action.\u003c/p\u003e \u003cp\u003eContemporary basketball research requires a dual-axis evaluation framework that integrates traditional game-related statistics with task-oriented performance indicators [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. Within this paradigm, play-type statistics have emerged as a transformative analytical approach, particularly due to their ability to describe pre-shot action executed through individual, paired, and team interactions [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. Recent studies have primarily focused on single or a limited set of specific actions-such as transitions [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e], cuts [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e], isolations [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e], and pick-and-roll plays [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e] across various levels of competitions, analyzing their frequencies and effectiveness within possessions. Findings consistently suggest that the increasing prevalence of pick-and-roll and transition team plays remains a defining trend, with high motion speed and cooperative actions proving generally more effective than individual ball-handler\u0026rsquo;s maneuvers [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. Marmarinos et al.'s [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e] analysis of 12,376 pick-and-roll sequences in Euroleague basketball demonstrated optimal efficiency under two conditions: (a) when a shot was attempted following two passes from the pick-and-roll action, and (b) when the screener successfully finished at the basket after rolling. Subsequent analyses of the 2020\u0026ndash;2021 men\u0026rsquo;s Euroleague season revealed that the most frequently used possessions and highest point-producing actions were catch-and-shoot plays and pick-and-roll ball handler possessions [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. These findings highlight the importance of spatial dynamic, well-coordinated team movements, and highly skilled players in enhancing play execution success in professional basketball. The coach\u0026rsquo;s decision to elect a particular play type in critical game moments, such as endgame situations, can be a decisive factor in a team\u0026rsquo;s win or loss [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. However, limited studies have integrated play-type statistics to predict overall team success [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. Conte et al. [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e] found that NCAA teams performing more hand-offs, but fewer off-ball screens were more likely to outscore their opponents. Similarly, Bustamante et al. [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e] identified higher successful rates of catch-and-shoot plays, cuts, and pick-and-roll (both handler and roller actions) among winning teams in the NBA 2019\u0026ndash;2020 season. While these metrics effectively capture certain aspects of performance, they may not fully encapsulate the complexities inherent in professional basketball gameplay when analyzed solely through traditional linear techniques [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eMachine learning (ML) algorithms have emerged as powerful tools for predictive modeling in basketball analytics, enabling multivariate analysis of performance indicators, particularly in areas such as training regimens, injury patterns, play-style strategies, and short-, mid-, and long-term match performance evaluation [\u003cspan additionalcitationids=\"CR27\" citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. ML algorithms including decision trees (DT), support vector machines (SVM), random forest (RF), artificial neural network (ANN), and XGBoost, have demonstrated superior capability in modeling non-linear interactions within multivariate performance datasets [\u003cspan additionalcitationids=\"CR30 CR31\" citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]. For instance, Leicht et al. [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e] implemented a hybrid analytical framework integrating conditional inference classification trees with logistic regression to identify KPIs associated with Olympic basketball success. A multimodal analysis revealed field-goal percentage, defensive rebounds, steals and turnovers were statistically significant predictors across both linear and non-linear models. In an NBA case study, ensemble modeling combining DT and RF algorithms identified shooting efficiency as the primary KPI, followed by defensive rebounds and opponent turnovers for the Golden State Warriors [\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]. While ML algorithms have demonstrated predictive superiority, model interpretability limitations persist, particularly in quantifying individual indicator contributions to outcome predictions [\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e]. To address this, Ouyang et al. [\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e] developed a SHAP-XGBoost framework integrating seven ML algorithms (KNN, LightGBM, SVM, RF, Logistic Regression, XGBoost and DT) with SHapley Additive exPlanations (SHAP). The findings identified that field goal percentage, defensive rebounds and turnovers as real-time predictors of NBA game outcomes.\u003c/p\u003e \u003cp\u003eThe Chinese Basketball Association (CBA), Asia\u0026rsquo;s premier professional league, exhibits distinct characteristics in tactical implementations, team structures, and stylistic approaches, differentiating it from other leagues [\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e, \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e]. Using basic game-related statistics, Cai et al. [\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e] developed a hybrid ensemble learning framework based on an SVM model to predict game outcomes during the 2016\u0026ndash;2017 regular season. Unlike the NBA and Euroleague, where spacing and three-point shooting have become increasingly dominant over the past decade [\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e, \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e], the CBA remains more reliant on mid-range shooting and paint-based plays, as reflected in its lower three-point attempt rate per game [\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e, \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e]. Additionally, studies indicate that the CBA employs fewer direct pick-and-roll plays compared to the NBA [\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e], which may influence the interpretation of KPIs in this context. These differences necessitate a specialized performance analysis approach, tailored to the distinct gameplay characteristics of the CBA.\u003c/p\u003e \u003cp\u003ePrevious studies have predominantly relied on traditional linear techniques to explore game-related KPIs, often neglecting comprehensive analyses that integrate both technical and tactical perspectives [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e, \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e]. Furthermore, prior ML-based basketball studies have primarily focused on single or hybrid models, with limited exploration of different ML algorithms to establish a comprehensive and stable set of KPIs across multi-season datasets [\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e, \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e]. This limitation affects the objectivity of game outcomes predictions and model interpretability, commonly referred to as \u0026ldquo;black box\u0026rdquo; issue, particularly in leagues outside the NBA and Euroleague [\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eTherefore, the novelty of our study lies in its focus on the CBA league and the application of multipliable ML classifiers for identifying critical KPIs and quantifying their impacts on game outcomes in a more comparative and interpretable manner. Firstly, traditional t-tests and K-means clustering methods are employed to preliminary selected game-related and play-type statistics associated with winning outcomes from the 2018\u0026ndash;2022 regular seasons. Subsequently, these parameters undergo essential pre-processing and are merged for feature selection using methods such as lasso and stepwise regression to confirm the final KPIs. Furthermore, ten different types of ML classifiers-including DT, RF, SVM, ANN, GBDT, XGBoost, Adaboost, CatBoost, LightGBM, and LR-are utilized to predict game outcomes based on the identified KPIs. Finally, the SHAP algorithm is implemented to quantify feature importance in the optimal-performing models, providing a more granular interpretation of KPI contributions to game outcomes.\u003c/p\u003e"},{"header":"2 Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1 Sample and variables\u003c/h2\u003e \u003cp\u003eThe data for this study were obtained from Hudl InStat Basketball, a commercially accessible provider based in the USA. A total of 3, 249 games from the 2018\u0026ndash;2022 CBA regular seasons were analyzed. The reliability and validity of the system have been verified [\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e]. To further assess the reliability of game performance data in the CBA, two experienced coaches were called up to analyze a randomly selected sample of ten games. The inter-observer kappa values and the kappa values of observed data and the data from Hudl InStat Basketball were all higher than 0.91. \u003cb\u003eTable\u0026nbsp;1\u003c/b\u003e presents the grouping of each parameter and the action definitions provided by InStat. Based on previous studies [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e], 56 basketball game performance parameters were selected and were categorized into two groups: 34 game-related and 22 play-type statistics.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Taba\" border=\"1\"\u003e \u003ccolgroup cols=\"2\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTable\u0026nbsp;1 Selected parameters and definitions\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eParameters\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDefinitions\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGame-related statistics\u003c/p\u003e \u003cp\u003eOffensive parameters\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eTwo-point field goals\u003c/b\u003e: field goals shots inside the three-point line or when the player\u0026rsquo;s foot is touching it\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eThree-point field goals\u003c/b\u003e: field goals attempted beyond the three-point line\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eFree throws\u003c/b\u003e: a shot from the free-throw line awarded after a foul on the shooter by the opposing team or after exceeding the foul limit\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eAssists\u003c/b\u003e: a pass to a teammate that directly leads to a made field goal\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eTurnovers\u003c/b\u003e: an action that results in the offensive team losing possession of the ball to the opposing team\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eFouls drawn\u003c/b\u003e: a foul committed by the opponents, stopping the game and followed by either a free throw\u003c/p\u003e \u003cp\u003e\u003cb\u003eOffensive rebounds\u003c/b\u003e: gaining control of the live ball by the offensive team after their missed shot attempt or free throw\u003c/p\u003e \u003cp\u003e\u003cb\u003eOffensive rebounds%\u003c/b\u003e: percentage of all available offensive rebounds a team can get. Formula\u0026thinsp;=\u0026thinsp;offensive rebounds / (offensive rebounds\u0026thinsp;+\u0026thinsp;opponent\u0026rsquo;s defensive rebounds)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDefensive parameters\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eSteals\u003c/b\u003e: an action where the defending player causes a turnover by taking away the ball\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eBlocks\u003c/b\u003e: a defensive action where a player legally deflects a field goal attempt by an offensive player\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eFouls\u003c/b\u003e: a breach of the rules, which can be categorized as personal, technical, unsportsmanlike, flagrant and team fouls\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eDefensive rebounds\u003c/b\u003e: gaining control of the live ball by the defensive team after a missed shot attempt or free throw of the opponent\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eDefensive rebounds%\u003c/b\u003e: percentage of all available defensive rebounds a team can get. Formula\u0026thinsp;=\u0026thinsp;defensive rebounds / (defensive rebounds\u0026thinsp;+\u0026thinsp;opponent\u0026rsquo;s offensive rebounds)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePlay-type statistics\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eTransitions\u003c/b\u003e: ball possessions that start at team\u0026rsquo;s backcourt and last between 4 and 8 seconds\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eCatch and shoot\u003c/b\u003e: a play that ends with a jump shot taken at least 3 meters from the rim by a player who controlled the ball less than 2 seconds or did not dribble\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eCatch and drive\u003c/b\u003e: a play that ends with a shot taken from 3 meters or less from the rim by a player who dribbled the ball after receiving a pass\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eDrive with shot\u003c/b\u003e: a play that ends with a shot taken from 3 meters or less from the rim by a player who dribbled\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003ePost up\u003c/b\u003e: an offensive player receives the ball within 4.5 meters of the basket with his back to it and attempts a shot\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eIsolations\u003c/b\u003e: a play in which a team gives the ball handler space to play one-on-one against his opponent\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eCuts\u003c/b\u003e: a play in which a player attempting to shoot receives the ball while running toward the rim. This includes screens, fast breaks, and situations where a player gets open near the rim\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eScreens-off\u003c/b\u003e: a player without the ball legally uses their body to block the defender\u0026rsquo;s movement, allowing a teammate to break free from the defense\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eOn-ball screens\u003c/b\u003e: a player sets a screen (pick) for a teammate handling the ball and then slips into the rim (rolls). There are three types: Pick and roll handler - when the ball handler makes a shot attempt; Pick and roll roller - when the screener makes a shot; pick and pop - when the screener stays away from the basket or remains stationary to create a shooting opportunity\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eContested shot\u003c/b\u003e: there is opponent between the rim and a shooter\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eUncontested shot\u003c/b\u003e: there is no opponent between the rim and a shooter\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2 Statistical analysis\u003c/h2\u003e \u003cp\u003e \u003cb\u003eStep 1\u003c/b\u003e: A k-means cluster analysis was applied to divide all the games into two groups based on the final score differences: balanced and unbalanced games. Following the inclusion criteria for balanced games, 952 unbalanced games defined as those with final score differences of 19 points or more (considered outliers in this context), were excluded from the analysis. Consequently, the final dataset included 674 games from the 2018\u0026ndash;2019 season (246 games were missing due to technical issues with the data provider), 920 games from the 2019\u0026ndash;2020 season, 1, 008 games from the 2020\u0026ndash;2021 season, and 760 games from the 2021\u0026ndash;2022 season.\u003c/p\u003e \u003cp\u003e \u003cb\u003eStep 2\u003c/b\u003e: Descriptive statistics (mean and standard deviation) were calculated for each game performance parameter in relation to game outcome (win/loss) for balanced games. An independent samples t-test was conducted to examine differences in each statistic between winning and losing teams. A Pearson\u0026rsquo;s correlation matrix was then constructed to evaluate collinearity among preliminary performance parameters. Correlation was categorized as follows: negligible or small (|r|\u0026lt;0.1), weak (0.1\u0026lt;|r|\u0026lt;0.3), moderate (0.3\u0026lt;|r|\u0026lt;0.5), strong (0.5\u0026lt;|r|\u0026lt;0.7), very strong (0.7\u0026lt;|r|\u0026lt;0.9), and extremely strong (|r|\u0026gt;0.9).\u003c/p\u003e \u003cp\u003e \u003cstrong\u003eStep 3\u003c/strong\u003e \u003cp\u003eLasso and stepwise regression were employed to identify the optimal set of KPIs from a large pool of candidate parameters. These technical and tactical parameters were treated as independent variables, with game outcome coded as the response variable (win\u0026thinsp;=\u0026thinsp;1, loss\u0026thinsp;=\u0026thinsp;0). Lasso regression incorporates a regularization term (L1-norm) that shrinks some regression coefficients to zero, retaining only significant predictors. This approach effectively eliminated irrelevant variables and selected significant ones [\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e]. The lasso regression process utilized 10-fold cross-validation across 100 \u0026ldquo;lambda\u0026rdquo; values, calculating the mean square error (MSE) at each value. The model with the smallest MSE was identified, and a model within one standard error of the minimum MSE was selected [\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e]. Stepwise regression was subsequently applied to compare its feature selection results with those of lasso regression, focusing on the overlap of selected features. This method refined the model by incrementally adding statistically significant independent variables and automatically removing insignificant ones [\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e].\u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eStep 4\u003c/strong\u003e \u003cp\u003eFollowing the identification of influential KPIs, a variety of mainstream ML classifiers were applied, including DT, RF, SVM, ANN, GBDT, XGBoost, Adaboost, CatBoost, LightGBM, and LR. A stochastic validation strategy was adopted by splitting the data into an 80% training set and 20% testing set. A grid search method was employed for hyperparameter optimization, with the optimal parameter combination subsequently used to construct the ML models. \u0026zwnj;A nested 10-fold cross-validation approach was adopted to prevent overfitting.\u0026zwnj; The performance of these models was evaluated using metrics such as accuracy, recall, precision, F1-score, and the Area Under the Receiver Operating Characteristic Curve (AUC). Finally, the SHAP algorithm was introduced to improve the interpretability of the optimal models\u0026rsquo; output. This method computed Shapley values for each feature based on game theory, quantifying their individual contributions to the model\u0026rsquo;s predictions [48].\u003c/p\u003e \u003c/p\u003e \u003cp\u003eAll data analyses were conducted using R programming language (version 3.6.3), with the package of \u0026ldquo;caret\u0026rdquo; for developing machine learning algorithms. Statistical significance was set at p\u0026thinsp;\u0026lt;\u0026thinsp;0.05.\u003c/p\u003e \u003c/div\u003e"},{"header":"3 Results","content":"\u003cp\u003e\u003cstrong\u003eTable 2 and 3\u003c/strong\u003e present the results of the descriptive statistics and independent t-tests. Significant differences were found between winning and losing teams for all game-related statistics except for contested and uncontested shot attempts. Furthermore, substantial differences were observed in transitions (made and attempted), catch-and-shoot made, pick-and-pop (PnP) attempted, and eight additional pre-shot combinations between the two groups.\u003c/p\u003e\n\u003cdiv\u003e\n \u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"538\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"5\" valign=\"bottom\" style=\"width: 100%;\"\u003e\n \u003cp\u003eTable 2 Descriptive statistics of game-related parameters of winning and losing teams\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.5093%;\"\u003e\n \u003cp\u003eParameters\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003eWin（n=1149）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003eLose（n=1148）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e\u003cem\u003eT\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003eP\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.5093%;\"\u003e\n \u003cp\u003etwo FG made\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e28.5\u0026plusmn;5.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e26.2\u0026plusmn;4.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e-11.199\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.5093%;\"\u003e\n \u003cp\u003etwo FG missed\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e23.7\u0026plusmn;5.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e25.1\u0026plusmn;5.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e6.228\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.5093%;\"\u003e\n \u003cp\u003etwo FG attempted\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e52.2\u0026plusmn;7.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e51.2\u0026plusmn;7.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e-3.119\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.5093%;\"\u003e\n \u003cp\u003etwo FG %\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e54.7\u0026plusmn;7.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e51.2\u0026plusmn;6.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e-12.215\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.5093%;\"\u003e\n \u003cp\u003ethree FG made\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e10.3\u0026plusmn;3.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e9.5\u0026plusmn;3.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e-6.144\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.5093%;\"\u003e\n \u003cp\u003ethree FG missed\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e17.4\u0026plusmn;4.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e19.1\u0026plusmn;4.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e8.913\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.5093%;\"\u003e\n \u003cp\u003ethree FG attempted\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e27.7\u0026plusmn;6.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e28.6\u0026plusmn;6.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e3.184\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.5093%;\"\u003e\n \u003cp\u003ethree FG %\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e37.1\u0026plusmn;8.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e32.9\u0026plusmn;8.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e-11.715\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.5093%;\"\u003e\n \u003cp\u003efree throws made\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e19.8\u0026plusmn;6.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e17.8\u0026plusmn;6.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e-7.340\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.5093%;\"\u003e\n \u003cp\u003efree throws missed\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e6.4\u0026plusmn;3.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e6.1\u0026plusmn;3.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e-2.182\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.015\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.5093%;\"\u003e\n \u003cp\u003efree throws attempted\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e26.1\u0026plusmn;8.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e23.9\u0026plusmn;7.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e-6.867\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.5093%;\"\u003e\n \u003cp\u003efree throws %\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e75.6\u0026plusmn;9.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e74.5\u0026plusmn;10.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e-2.369\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.008\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.5093%;\"\u003e\n \u003cp\u003euncontested shot made\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e6.3\u0026plusmn;4.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e5.8\u0026plusmn;3.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e-2.972\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.5093%;\"\u003e\n \u003cp\u003euncontested shot missed\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e8.9\u0026plusmn;5.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e9.6\u0026plusmn;6.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e2.851\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.002\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.5093%;\"\u003e\n \u003cp\u003euncontested shot attempted\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e15.2\u0026plusmn;9.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e15.4\u0026plusmn;9.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e0.594\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e0.276\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.5093%;\"\u003e\n \u003cp\u003euncontested shot %\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e42.3\u0026plusmn;15.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e38.1\u0026plusmn;15.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e-6.411\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.5093%;\"\u003e\n \u003cp\u003econtested shot made\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e29.6\u0026plusmn;5.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e27.2\u0026plusmn;5.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e-10.723\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.5093%;\"\u003e\n \u003cp\u003econtested shot missed\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e32.2\u0026plusmn;7.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e34.6\u0026plusmn;7.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e7.646\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.5093%;\"\u003e\n \u003cp\u003econtested shot attempted\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e61.8\u0026plusmn;10.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e61.8\u0026plusmn;10.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e-0.099\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e0.461\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.5093%;\"\u003e\n \u003cp\u003econtested shot %\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e48.2\u0026plusmn;6.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e44.2\u0026plusmn;6.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e-14.708\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 27.5093%;\"\u003e\n \u003cp\u003eassists\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e22.4\u0026plusmn;4.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e19.7\u0026plusmn;4.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e-13.083\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 27.5093%;\"\u003e\n \u003cp\u003eturnovers\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e14.3\u0026plusmn;4.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e15.1\u0026plusmn;4.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e4.777\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 27.5093%;\"\u003e\n \u003cp\u003efouls drawn\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e24.3\u0026plusmn;4.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e23.3\u0026plusmn;4.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e-4.886\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 27.5093%;\"\u003e\n \u003cp\u003esteals\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e8.2\u0026plusmn;3.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e7.6\u0026plusmn;2.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e-4.799\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 27.5093%;\"\u003e\n \u003cp\u003eblocks\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e3.7\u0026plusmn;2.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e3.1\u0026plusmn;1.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e-6.283\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 27.5093%;\"\u003e\n \u003cp\u003efouls\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e23.3\u0026plusmn;4.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e24.3\u0026plusmn;4.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e4.910\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 27.5093%;\"\u003e\n \u003cp\u003eoffensive rebounds\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e11.8\u0026plusmn;4.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e11.2\u0026plusmn;4.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e-3.579\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.5093%;\"\u003e\n \u003cp\u003edefensive rebounds\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e30.3\u0026plusmn;4.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e27.1\u0026plusmn;4.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e-16.455\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.5093%;\"\u003e\n \u003cp\u003eoffensive rebounds %\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e30.0\u0026plusmn;8.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e26.7\u0026plusmn;7.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e-9.953\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.5093%;\"\u003e\n \u003cp\u003edefensive rebounds %\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.8885%;\"\u003e\n \u003cp\u003e73.3\u0026plusmn;7.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.0037%;\"\u003e\n \u003cp\u003e70.0\u0026plusmn;8.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.684%;\"\u003e\n \u003cp\u003e-9.911\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.9145%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cdiv\u003e\n \u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"533\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"5\" valign=\"bottom\" style=\"width: 100%;\"\u003e\n \u003cp\u003eTable 3 Descriptive statistics of play-type parameters of winning and losing teams\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.2045%;\"\u003e\n \u003cp\u003eParameters\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.5122%;\"\u003e\n \u003cp\u003eWin（n=1149）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.576%;\"\u003e\n \u003cp\u003eLose（n=1148）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.8218%;\"\u003e\n \u003cp\u003e\u003cem\u003eT\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.8856%;\"\u003e\n \u003cp\u003eP\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.2045%;\"\u003e\n \u003cp\u003etransitions made\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.5122%;\"\u003e\n \u003cp\u003e7.1\u0026plusmn;3.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.576%;\"\u003e\n \u003cp\u003e6.2\u0026plusmn;2.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.8218%;\"\u003e\n \u003cp\u003e-7.161\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.8856%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.2045%;\"\u003e\n \u003cp\u003etransitions attempted\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.5122%;\"\u003e\n \u003cp\u003e11.4\u0026plusmn;4.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.576%;\"\u003e\n \u003cp\u003e10.9\u0026plusmn;4.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.8218%;\"\u003e\n \u003cp\u003e-2.302\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.8856%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.021\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.2045%;\"\u003e\n \u003cp\u003ecatch and shot made\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.5122%;\"\u003e\n \u003cp\u003e6.5\u0026plusmn;2.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.576%;\"\u003e\n \u003cp\u003e5.9\u0026plusmn;2.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.8218%;\"\u003e\n \u003cp\u003e-5.232\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.8856%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.2045%;\"\u003e\n \u003cp\u003ecatch and shot attempted\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.5122%;\"\u003e\n \u003cp\u003e17\u0026plusmn;5.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.576%;\"\u003e\n \u003cp\u003e16.7\u0026plusmn;5.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.8218%;\"\u003e\n \u003cp\u003e-1.236\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.8856%;\"\u003e\n \u003cp\u003e0.217\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.2045%;\"\u003e\n \u003cp\u003ecatch and drive made\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.5122%;\"\u003e\n \u003cp\u003e2.8\u0026plusmn;1.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.576%;\"\u003e\n \u003cp\u003e2.8\u0026plusmn;1.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.8218%;\"\u003e\n \u003cp\u003e-1.026\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.8856%;\"\u003e\n \u003cp\u003e0.305\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.2045%;\"\u003e\n \u003cp\u003ecatch and drive attempted\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.5122%;\"\u003e\n \u003cp\u003e6.7\u0026plusmn;3.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.576%;\"\u003e\n \u003cp\u003e7.1\u0026plusmn;3.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.8218%;\"\u003e\n \u003cp\u003e3.221\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.8856%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.2045%;\"\u003e\n \u003cp\u003edrive with shot made\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.5122%;\"\u003e\n \u003cp\u003e7.2\u0026plusmn;3.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.576%;\"\u003e\n \u003cp\u003e7.1\u0026plusmn;3.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.8218%;\"\u003e\n \u003cp\u003e-0.708\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.8856%;\"\u003e\n \u003cp\u003e0.479\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.2045%;\"\u003e\n \u003cp\u003edrive with shot attempted\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.5122%;\"\u003e\n \u003cp\u003e14.8\u0026plusmn;5.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.576%;\"\u003e\n \u003cp\u003e15.7\u0026plusmn;6.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.8218%;\"\u003e\n \u003cp\u003e3.379\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.8856%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.2045%;\"\u003e\n \u003cp\u003epost up made\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.5122%;\"\u003e\n \u003cp\u003e2.2\u0026plusmn;2.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.576%;\"\u003e\n \u003cp\u003e2.2\u0026plusmn;2.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.8218%;\"\u003e\n \u003cp\u003e-0.887\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.8856%;\"\u003e\n \u003cp\u003e0.375\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.2045%;\"\u003e\n \u003cp\u003epost up attempted\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.5122%;\"\u003e\n \u003cp\u003e4.8\u0026plusmn;3.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.576%;\"\u003e\n \u003cp\u003e4.9\u0026plusmn;3.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.8218%;\"\u003e\n \u003cp\u003e1.318\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.8856%;\"\u003e\n \u003cp\u003e0.188\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.2045%;\"\u003e\n \u003cp\u003eisolation made\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.5122%;\"\u003e\n \u003cp\u003e2.3\u0026plusmn;2.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.576%;\"\u003e\n \u003cp\u003e2.2\u0026plusmn;2.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.8218%;\"\u003e\n \u003cp\u003e-1.252\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.8856%;\"\u003e\n \u003cp\u003e0.211\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.2045%;\"\u003e\n \u003cp\u003eisolation attempted\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.5122%;\"\u003e\n \u003cp\u003e5.4\u0026plusmn;4.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.576%;\"\u003e\n \u003cp\u003e5.6\u0026plusmn;4.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.8218%;\"\u003e\n \u003cp\u003e1.027\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.8856%;\"\u003e\n \u003cp\u003e0.304\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.2045%;\"\u003e\n \u003cp\u003ecuts made\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.5122%;\"\u003e\n \u003cp\u003e4.3\u0026plusmn;2.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.576%;\"\u003e\n \u003cp\u003e3.6\u0026plusmn;2.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.8218%;\"\u003e\n \u003cp\u003e-6.463\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.8856%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.2045%;\"\u003e\n \u003cp\u003ecuts attempted\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.5122%;\"\u003e\n \u003cp\u003e6.4\u0026plusmn;3.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.576%;\"\u003e\n \u003cp\u003e5.6\u0026plusmn;3.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.8218%;\"\u003e\n \u003cp\u003e-6.029\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.8856%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.2045%;\"\u003e\n \u003cp\u003escreens off made\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.5122%;\"\u003e\n \u003cp\u003e1.9\u0026plusmn;1.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.576%;\"\u003e\n \u003cp\u003e1.6\u0026plusmn;1.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.8218%;\"\u003e\n \u003cp\u003e-3.901\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.8856%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.2045%;\"\u003e\n \u003cp\u003escreens off attempted\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.5122%;\"\u003e\n \u003cp\u003e4.1\u0026plusmn;2.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.576%;\"\u003e\n \u003cp\u003e4.0\u0026plusmn;2.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.8218%;\"\u003e\n \u003cp\u003e-0.665\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.8856%;\"\u003e\n \u003cp\u003e0.506\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.2045%;\"\u003e\n \u003cp\u003ePnR handlers made\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.5122%;\"\u003e\n \u003cp\u003e4.8\u0026plusmn;2.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.576%;\"\u003e\n \u003cp\u003e4.4\u0026plusmn;2.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.8218%;\"\u003e\n \u003cp\u003e-2.921\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.8856%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.004\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.2045%;\"\u003e\n \u003cp\u003ePnR handlers attempted\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.5122%;\"\u003e\n \u003cp\u003e11.1\u0026plusmn;5.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.576%;\"\u003e\n \u003cp\u003e11\u0026plusmn;5.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.8218%;\"\u003e\n \u003cp\u003e0.369\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.8856%;\"\u003e\n \u003cp\u003e0.712\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.2045%;\"\u003e\n \u003cp\u003ePnR rollers made\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.5122%;\"\u003e\n \u003cp\u003e1.3\u0026plusmn;1.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.576%;\"\u003e\n \u003cp\u003e1.2\u0026plusmn;1.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.8218%;\"\u003e\n \u003cp\u003e-2.572\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.8856%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.010\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.2045%;\"\u003e\n \u003cp\u003ePnR rollers attempted\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.5122%;\"\u003e\n \u003cp\u003e2.1\u0026plusmn;1.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.576%;\"\u003e\n \u003cp\u003e2.0\u0026plusmn;1.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.8218%;\"\u003e\n \u003cp\u003e-1.369\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.8856%;\"\u003e\n \u003cp\u003e0.163\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.2045%;\"\u003e\n \u003cp\u003ePnP made\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.5122%;\"\u003e\n \u003cp\u003e0.8\u0026plusmn;1.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.576%;\"\u003e\n \u003cp\u003e0.9\u0026plusmn;1.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.8218%;\"\u003e\n \u003cp\u003e0.829\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.8856%;\"\u003e\n \u003cp\u003e0.407\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 27.2045%;\"\u003e\n \u003cp\u003ePnP attempted\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 19.5122%;\"\u003e\n \u003cp\u003e2.0\u0026plusmn;1.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 21.576%;\"\u003e\n \u003cp\u003e2.2\u0026plusmn;2.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 14.8218%;\"\u003e\n \u003cp\u003e2.531\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.8856%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.011\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003e\u003cstrong\u003eFigure 1\u003c/strong\u003e illustrates the relationships between the preliminary performance parameters. Strong relationships were observed between free throws made and free throws attempted, offensive rebounds and offensive rebounds%, transitions made and transitions attempted, as well as free throws attempted and fouls drawn. Parameters related to two-point field goals demonstrated very strong correlations.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFigure 2a\u003c/strong\u003e displays the lasso regression path, showing how feature selection was achieved by adjusting the regularization parameter. \u003cstrong\u003eFigure 2b\u003c/strong\u003e presents the cross-validation results for lasso regression. By optimizing the regularization parameter \u0026lambda; through ten-fold cross-validation, the model identified the optimal log(\u0026lambda;) value at approximately -4.69 based on the 1-SE rule, selecting a total of 23 parameters. During the process, coefficients for 16 performance parameters such as two FG made and attempted, three FG made and attempted, free throw attempted and missed, contested shot made and missed, uncontested shot made and percentage, offensive rebounds, transitions made, cuts made and attempted, catch-and-drive attempted, PnR rollers made, were shrunk to zero and excluded.\u0026nbsp;\u003c/p\u003e\n\u003cdiv\u003e\n \u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"499\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"5\" style=\"width: 499px;\"\u003e\n \u003cp\u003eTable 3 Identification of key performance indicators based on lasso+logit regression\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 137px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003eEstimate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003eSE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003eZ\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003eP\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 137px;\"\u003e\n \u003cp\u003etwo FG missed\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e-0.21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.03\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e-7.16\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 137px;\"\u003e\n \u003cp\u003etwo FG%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.02\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e-0.20\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e0.844\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 137px;\"\u003e\n \u003cp\u003ethree FG missed\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e-0.21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.03\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e-8.14\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 137px;\"\u003e\n \u003cp\u003ethree FG %\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.06\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e4.43\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 137px;\"\u003e\n \u003cp\u003efree throws made\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.02\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e0.50\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e0.616\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 137px;\"\u003e\n \u003cp\u003efree throws%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.02\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e2.45\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.014\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 137px;\"\u003e\n \u003cp\u003eassists\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.02\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e3.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.003\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 137px;\"\u003e\n \u003cp\u003eturnovers\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e-0.34\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.02\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e-15.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 137px;\"\u003e\n \u003cp\u003efouls drawn\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.08\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.02\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e3.38\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 137px;\"\u003e\n \u003cp\u003euncontested shot missed\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e-0.02\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e-1.43\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e0.153\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 137px;\"\u003e\n \u003cp\u003econtested shot%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.03\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.02\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e1.42\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e0.157\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 137px;\"\u003e\n \u003cp\u003esteals\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.43\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.03\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e15.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 137px;\"\u003e\n \u003cp\u003eblocks\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.14\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.03\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e4.50\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 137px;\"\u003e\n \u003cp\u003efouls\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e-0.06\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e-3.94\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 137px;\"\u003e\n \u003cp\u003edefensive rebounds\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.34\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.02\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e17.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 137px;\"\u003e\n \u003cp\u003eoffensive rebound%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e15.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 137px;\"\u003e\n \u003cp\u003edefensive rebound%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e1.69\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e0.091\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 137px;\"\u003e\n \u003cp\u003etransitions attempted\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e-0.03\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.02\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e-1.80\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e0.071\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 137px;\"\u003e\n \u003cp\u003ecatch and shot made\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.06\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.03\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e1.92\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e0.055\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 137px;\"\u003e\n \u003cp\u003edrive with shot\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e-0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e-1.21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e0.225\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 137px;\"\u003e\n \u003cp\u003escreens off made\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.14\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.04\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e3.15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.002\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 137px;\"\u003e\n \u003cp\u003ePnR handlers made\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.08\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.03\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e3.12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.002\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 137px;\"\u003e\n \u003cp\u003ePnP attempted\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.07\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 96px;\"\u003e\n \u003cp\u003e0.03\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 100px;\"\u003e\n \u003cp\u003e2.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.040\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"90%\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"5\" style=\"width: 100px;\"\u003e\n \u003cp\u003eTable 4 Identification of key performance indicators based on stepwise regression\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 27px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 14px;\"\u003e\n \u003cp\u003eEstimate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003eSE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003eZ\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003eP\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 27px;\"\u003e\n \u003cp\u003etwo FG missed\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 14px;\"\u003e\n \u003cp\u003e-0.231\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e0.016\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e-14.236\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 27px;\"\u003e\n \u003cp\u003ethree FG attempted\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 14px;\"\u003e\n \u003cp\u003e0.160\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e0.021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e7.642\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 27px;\"\u003e\n \u003cp\u003ethree FG missed\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 14px;\"\u003e\n \u003cp\u003e-0.457\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e0.032\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e-14.292\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 27px;\"\u003e\n \u003cp\u003efree throws missed\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 14px;\"\u003e\n \u003cp\u003e-0.091\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e0.024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e-3.885\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 27px;\"\u003e\n \u003cp\u003eassists\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 14px;\"\u003e\n \u003cp\u003e0.060\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e0.015\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e4.062\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 27px;\"\u003e\n \u003cp\u003eturnovers\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 14px;\"\u003e\n \u003cp\u003e-0.350\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e0.021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e-16.657\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 27px;\"\u003e\n \u003cp\u003efouls drawn\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 14px;\"\u003e\n \u003cp\u003e0.115\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e0.017\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e6.640\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 27px;\"\u003e\n \u003cp\u003esteals\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 14px;\"\u003e\n \u003cp\u003e0.428\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e0.027\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e15.730\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 27px;\"\u003e\n \u003cp\u003eblocks\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 14px;\"\u003e\n \u003cp\u003e0.130\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e0.030\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e4.357\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 27px;\"\u003e\n \u003cp\u003efouls\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 14px;\"\u003e\n \u003cp\u003e-0.059\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e0.015\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e-3.985\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 27px;\"\u003e\n \u003cp\u003edefensive rebounds\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 14px;\"\u003e\n \u003cp\u003e0.354\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e0.018\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e19.557\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 27px;\"\u003e\n \u003cp\u003eoffensive rebounds%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 14px;\"\u003e\n \u003cp\u003e0.158\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e0.010\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e16.216\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 27px;\"\u003e\n \u003cp\u003etransitions made\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 14px;\"\u003e\n \u003cp\u003e0.101\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e0.039\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e2.583\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.010\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 27px;\"\u003e\n \u003cp\u003etransitions attempted\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 14px;\"\u003e\n \u003cp\u003e-0.089\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e0.027\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e-3.343\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 27px;\"\u003e\n \u003cp\u003eScreens-off made\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 14px;\"\u003e\n \u003cp\u003e0.118\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e0.041\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e2.881\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.004\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 27px;\"\u003e\n \u003cp\u003ePnR handlers made\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 14px;\"\u003e\n \u003cp\u003e0.066\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e0.023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19px;\"\u003e\n \u003cp\u003e2.805\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 18px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.005\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cem\u003eNote: only statistically significant indicators (p \u0026lt; 0.05) are presented in the table\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eOf the 23 parameters retained through lasso regression, defensive rebounds, offensive rebounds%, three FG%, and free throw% emerged as key predictors of CBA game outcomes. Two FG missed, three FG missed, turnovers and fouls, had significant negative impacts, while all other indicators positively influenced game outcomes (\u003cstrong\u003eTable 3\u003c/strong\u003e). In comparison, stepwise regression identified 16 significant performance indicators, with transitions attempted showing a significant negative impact (\u003cstrong\u003eTable 4\u003c/strong\u003e).\u003c/p\u003e\n\u003cp\u003eA comparative analysis of lasso and stepwise regression revealed a common subset of KPIs selected by both methods (\u003cstrong\u003eFigure 3\u003c/strong\u003e). These included two FG missed, three FG missed,\u0026nbsp;assists, turnovers,\u0026nbsp;fouls drawn, and\u0026nbsp;offensive rebounds% (offensive parameters); steals, blocks, fouls and defensive rebounds (defensive parameters) from game-related statistics; as well as\u0026nbsp;screens-off made and PnR handlers made\u0026nbsp;from pre-shot combinations.\u003c/p\u003e\n\u003cdiv\u003e\n \u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"6\" valign=\"top\" style=\"width: 544px;\"\u003e\n \u003cp\u003eTable 5 Performance evaluation metrics for the employed ML classifiers\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\u0026nbsp;\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003eAccuracy\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003ePrecision\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003eRecall\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003eF1-score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003eAUC\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eDT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e65.7%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e63.2%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e62.6%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.629\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003e0.703\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eRF\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e76.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e73.9%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e75.2%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.745\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003e0.852\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eSVM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e82.0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e80.8%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e80.4%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.806\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003e0.913\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eANN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e83.3%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e81.9%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e82.2%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.821\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003e0.914\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eXGBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e82.4%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e81.5%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e80.4%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.809\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003e0.897\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eGBDT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e81.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e79.3%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e80.4%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.798\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003e0.884\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eAdaboost\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e78.9%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e78.0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e76.2%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.771\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003e0.880\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eCatBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e81.3%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e79.4%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e80.8%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.801\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003e0.895\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eLightGBM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e79.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e77.8%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e77.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.775\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003e0.889\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003eLR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e81.3%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e79.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e81.3%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.802\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003e0.899\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003e\u003cstrong\u003eTable 5\u003c/strong\u003e summarizes the performance metrics of the ten machine learning algorithms employed in this study. Hyperparameter optimization was conducted via grid search, and ten-fold cross-validation was performed to evaluate classifier performance on the test set, ensuring result reliability and robustness. Among the classifiers, ANN achieved the highest accuracy at 83.3%, followed by XGBoost (82.4%) and SVM (82.0%). The DT classifier exhibited the lowest accuracy (65.7%), suggesting its limitations in capturing complex patterns. ANN also achieved the highest F1-score (0.821), reflecting an optimal balance between precision and recall. Furthermore, ANN achieved the highest AUC (0.914), closely followed by SVM with an AUC of 0.913. Overall, these results identified ANN, SVM, XGBoost, and LR as the top-performing classifiers for binary classification task.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFigure 4\u003c/strong\u003e illustrates the influence of the most informative KPIs on the predictions of the top-performing classifiers. SHAP value quantified the contribution of each feature, revealing that turnovers, two and three FG missed, and fouls were strongly associated with losing probability. Conversely, defensive rebounds, offensive rebounds percentage, assists, fouls drawn, steals, blocks, screens-off made, and PnR handlers made were positively associated with winning probability.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFigure 5\u003c/strong\u003e compares the feature importance rankings across four ML classifiers (ANN, SVM, XGBoost, and LR). Defensive rebounds (DREB) consistently ranked as the most critical KPI across all classifiers, underscoring their predictive value for game outcomes. Offensive rebounds percentage (OREB%) also ranked among the top three, particularly in ANN and LR. Additionally, missed two FG (2P_Missed) showed higher importance in SVM, XGBoost, and LR, but ranked slightly lower in ANN. Turnovers (TOV) ranked third in ANN but showed reduced importance in other classifiers. Steals (STL), a defensive KPI, ranked in the mid-range for all classifiers, while offensive KPIs like assists (AST) and fouls drawn (FD) appeared in the mid-to-lower range. Defensive indicators such as fouls (FOUL) and blocks (BLK), along with pre-shot combinations like PnR handlers made (PnR_H_Made) and screens-off made（SO_Made）, consistently ranked lower across all classifiers, however, their inclusion highlights their nuanced contribution to predictive frameworks, offering deeper insights into game dynamics.\u003c/p\u003e"},{"header":"4 Discussion","content":"\u003cp\u003eThis study advances basketball performance analytics by systematically identifying KPIs in the CBA through different ML classifiers while contextualizing the findings within the broader context of global basketball leagues. \u0026zwnj;Three primary contributions emerge\u0026zwnj;: (1) the identification of CBA-specific KPIs that differ from those in the NBA and European leagues, (2) the integration of tactical parameters (e.g., off- and on-ball screen execution) with traditional game statistics to enhance predictive accuracy, and (3) methodological insights into ML classifier interpretability for quantifying the most predictive indicators of CBA game outcomes and assessing their relative importance. While defensive rebounding and shooting efficiency have been widely recognized as KPIs in professional basketball (NBA, Euroleague), our study reveals that in the CBA, missed two-point attempts have a significantly higher impact on game outcomes than other shooting metrics. This suggests that shot selection dynamics differ in the CBA, with teams placing greater emphasis on mid-range efficiency rather than high-volume three-point shooting. Additionally, our results show that successful pick-and-roll handler plays have a lower predictive value in the CBA than in the NBA, potentially reflecting differences in tactical execution and spacing effectiveness within this league.\u003c/p\u003e \u003cp\u003eThe prominence of \u0026zwnj;defensive rebounds\u0026zwnj; and offensive rebound percentage\u0026zwnj; as top KPIs aligns with prior studies, reinforcing the universal significance of securing possession after missed shots as a key driver of success [7,49]. The predictive power of defensive rebounding aligns with findings from the NBA, ACB, EuroLeague, and Olympic competitions, underscoring its crucial role in determining professional basketball results [6,12,25,30,50]. Interestingly, while winning teams demonstrated a higher offensive rebounds percentage, they did not necessarily record a greater absolute number of offensive rebounds. This highlights the value of offensive rebounds percentage as a more advanced metric that captures rebounding efficiency rather than just the total count [51]. Similarly, Cs\u0026aacute;taljay et al. [52] found that offensive rebounds led to significantly higher shooting efficiency and an increased frequency of shooting fouls compared to possession changes, reinforcing the pivotal role of this metric. However, this finding diverges from NBA trends, where offensive rebounds percentage becomes particularly crucial in the later phases of games for winning teams [53]. This suggests that CBA teams may prioritize positioning for high probability putbacks rather than aggressively pursuing every offensive rebound, thereby reducing exposure to transition vulnerabilities. Additionally, the focus of this study on balanced games may account for some discrepancies with from prior research [3,52]. Balanced games typically exhibit different dynamics compared to lopsided contests, potentially altering the relative importance of indicators such as offensive rebounds percentage [24].\u003c/p\u003e \u003cp\u003eNotably, the methodological integration of \u0026zwnj;lasso regression\u0026zwnj;, \u0026zwnj;stepwise selection\u0026zwnj;, and \u0026zwnj;SHAP-driven ML interpretability\u0026zwnj; addresses a critical gap in basketball analytics. For example, three FG% and free throws% emerged as significant indicators in lasso regression but not in stepwise regression. These discrepancies highlight the importance of employing complementary methods to effectively identify KPIs from a large pool of candidate variables [43]. Traditional studies often rely on single league dataset or linear model, limiting the generalizability of findings [7,24]. By contrast, our multi-classifier framework including ANN, SVM, XGBoost, LR revealed that \u0026zwnj;missed two-point attempts\u0026zwnj;-a metric rarely prioritized in NBA analyses-were more predictive of CBA outcomes than successful shots. This finding underscores the \u0026zwnj;CBA\u0026rsquo;s unique shot selection dynamics\u0026zwnj;: teams with fewer missed two-point attempts likely demonstrate superior decision-making in spatially constrained offensive systems, a pattern less pronounced in leagues with higher three-point utilization seen in ACB and NBA leagues [49]. These insights validate the need for league-specific modeling, as KPIs derived from NBA data may misrepresent tactical priorities in Asian basketball contexts.\u003c/p\u003e \u003cp\u003eOther game-related statistics, such as blocks, turnovers, and fouls, also exhibited associations with winning likelihood, although these KPIs have not always been consistently emphasized [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]. While blocks and fouls ranked relative lower in importance, Sampaio et al. [\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e] reported that, in addition to shooting performance, recovered possessions-such as steals, blocks, and fewer turnovers-were among the primary contributors to point differentials during the 2008 Beijing Olympic Games. The ability to recover possession and convert it into effective scoring opportunities was a hallmark of team USA\u0026rsquo;s dominance in this tournament. The paradoxical role of fouls in CBA outcomes highlights cultural and tactical distinctions [\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e]. While losing teams committed more defensive fouls, a pattern consistent with NBA trends [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e], the weaker predictive value of fouls in determining loss likelihood partly suggests that the \u0026zwnj;strategic fouling is less systematically employed\u0026zwnj; in the CBA compared to elite international competitions [55]. Although fouls provide opponents with free-throw opportunities, they may also indicate intense defensive pressure [56]. Strategic fouling can serve as an effective tactic for disrupting the opponent\u0026rsquo;s rhythm, limiting fast-break opportunities, and neutralizing tactical setups [57]. However, in the CBA, the lower reliance on intentional fouling in late-game scenarios suggests a distinct coaching philosophy that prioritizes maintaining game tempo over disrupting play. This difference may stem from CBA coaching principles favoring a faster-paced offensive transitions, thereby reducing the frequency of \u0026ldquo;foul-to-stop-clock\u0026rdquo; strategies [29]. Similarly, the limited predictive power of blocks contrasts with their significance in the NBA, where elite rim protectors exert a disproportionate influence on game outcomes [50]. This divergence may be attributed to the CBA\u0026rsquo;s emphasis on team-oriented defense rather than individual shot-blocking prowess\u0026zwnj;, a characteristic that aligns with broader trends in Asian basketball development philosophies.\u003c/p\u003e \u003cp\u003eThis study also examined tactical parameters critical to team success alongside game-related statistics. Among these, only ball screen play-types significantly increased the likelihood of winning, underscoring the importance of this commonly used tactical behavior in high-level basketball competitions [\u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e58\u003c/span\u003e]. This study further emphasizes the significance of executing more PnR handler\u0026rsquo;s successful shots for success compared to other on-ball screen types (e.g., roll-in or pop-out), aligning with findings from NBA studies [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. However, this result contrasts with Bustamante et al. [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e], who found that PnR handler made plays had negligible predictive value in NBA games. Variations in offensive strategies may explain these differences. For instance, coaches frequently encourage players to attack the basket aggressively after executing a PnR, making the ball handler\u0026rsquo;s shot a central component of this tactic [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. This strategy is particularly evident in teams with elite individual scorers, who often rely on isolation-based finishes to capitalize on one-on-one advantages [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. These advantages become particularly valuable in high-pressure situations, such as when structured team play breaks down or the shot clock winds down [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eWhile ball screens are widely recognized as a cornerstone of modern offenses in professional basketball competitions [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e], their lower predictive value in this study highlights unique tactical challenges in the CBA. Specifically, off-ball screen sequences-which require precise coordination between passers, cutters, and screeners to generate uncontested shots-appear to be either underutilized or inefficiently executed in CBA gameplay [\u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e59\u003c/span\u003e]. Unlike in European-style offenses, where high-IQ playmakers and sharpshooters excel in reading defensive rotations and exploiting screens to create open looks for elite perimeter shooters [\u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e60\u003c/span\u003e], CBA teams may face challenges in spacing, execution speed, and defensive adaptations, which limit the overall impact of these actions. This is evidenced by the low success rate of off-ball screens, even among winning teams. Furthermore, variations in defensive schemes, such as the frequent use of switching defenses, may further reduce opportunities for executing multi-step off-ball actions, forcing teams to rely more on individual shot creation [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThese results also reinforce the notion that it is not merely the quantity of successful screens but rather their finishing efficiency that determines team success [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. The similar shooting performance of CBA teams in both contested and uncontested scenarios suggests that screening actions do not consistently generate a significant scoring advantage. This inefficiency diminishes the tactical payoff of screening efforts, rendering them statistically less impactful [\u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e59\u003c/span\u003e]. Finally, the limited importance of \u0026zwnj;successful off-ball screens and PnR handler plays \u0026zwnj;in CBA outcomes highlights a critical divergence from global basketball trends. As Marmarinos et al. [17] observed, cooperative actions among players, particularly those involving multiple interconnected plays, are generally more effective than isolated actions by ball handlers. This suggests that CBA teams may benefit from enhancing offensive sequences that focus on greater synergy and continuity, enabling additional scoring opportunities for teammates in optimal shooting positions.\u003c/p\u003e"},{"header":"5 Conclusion and future research lines","content":"\u003cp\u003eThis study aimed to systematically identify KPIs in the CBA using various machine learning classifiers. By integrating traditional game-related statistics with tactical play-type metrics, the study provided data-driven insights into winning strategies in the CBA. The findings highlight the importance of possession retention through rebounding and effective two-point shot selection as key predictors of game outcomes. This study provides a roadmap for integrating ML-driven performance analysis into team decision-making processes. Additionally, automated scouting reports, based on based on identified CBA-specific performance indicators, enable coaching staff to optimize player roles, adjust rotation strategies, and refine opponent-specific preparations.\u003c/p\u003e \u003cp\u003eFuture studies should expand on these findings by examining ML explainability in live game scenarios, thereby enhancing their practical usability for coaching applications. Specifically, incorporating temporal sequence modeling (e.g., recurrent neural networks) could capture in-game momentum shifts and offer deeper insights into how KPIs evolve over different game phases. Another key direction is the integration of ML models into coaching software. When combined with markerless optical tracking systems, teams can automate possession-based performance analysis and evaluate the impact of off-ball movements and pick-and-roll strategies. Finally, a cross-league comparative analysis incorporating data from the NBA, Euroleague, and other professional leagues, could provide a broader validation of CBA-specific KPIs, contributing to the development of universal framework for performance optimization in professional basketball.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll data generated or analysed during this study are included in this published article [and its supplementary information files].\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors have no conflicts of interest to declare that are relevant to the content of this article.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe author received no financial support for the research.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors' contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRD: Conceptualization, Investigation, Validation, Visualization, Writing – original draft. MG: Methodology, Formal analysis, Writing – review \u0026amp; editing. SZ: Methodology, Formal analysis, Writing – review \u0026amp; editing. XX: Software, Writing – review \u0026amp; editing. ZL: Data curation, Writing – review \u0026amp; editing. BY: Visualization, Writing – review \u0026amp; editing. HL: Conceptualization, Project administration, Resources, Writing – review \u0026amp; editing.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eComments and suggestions from the anonymous reviewers which improved the quality of this paper are appreciated.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eAkinci Y. Examining the Differences Between Playoff Teams and Non-Playoff Teams in Men\u0026rsquo;s Euroleague; Play-Type Statistics Perspective. Sage Open. 2023;13(4):21582440231220155. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1177/21582440231220155\u003c/span\u003e\u003cspan address=\"10.1177/21582440231220155\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eG\u0026oacute;mez M\u0026Aacute;, Battaglia O, Lorenzo A, Lorenzo J, Jim\u0026eacute;nez S, Sampaio J. Effectiveness during ball screens in elite basketball games. J Sports Sci. 2015;33(17):1844\u0026ndash;52. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/02640414.2015.1014829\u003c/span\u003e\u003cspan address=\"10.1080/02640414.2015.1014829\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eConte D, Tessitore A, Gjullin A, Mackinnon D, Lupo C, Favero T. Investigating the game-related statistics and tactical profile in NCAA division I men\u0026rsquo;s basketball games. Biol Sport. 2018;35(2):137\u0026ndash;43. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.5114/biolsport.2018.71602\u003c/span\u003e\u003cspan address=\"10.5114/biolsport.2018.71602\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChristmann J, Akamphuber M, M\u0026uuml;llenbach AL, G\u0026uuml;llich A. Crunch time in the NBA \u0026ndash; The effectiveness of different play types in the endgame of close matches in professional basketball. Int J Sports Sci Coaching. 2018;13(6):1090\u0026ndash;9. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1177/1747954118772485\u003c/span\u003e\u003cspan address=\"10.1177/1747954118772485\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhou W, Sansone P, Jia Z, Gomez MA, Li F. Determining the key performance indicators on game outcomes in NBA based on quantile regression analysis. Int J Perform Anal Sport. 2024;0(0):1\u0026ndash;16. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/24748668.2024.2325846\u003c/span\u003e\u003cspan address=\"10.1080/24748668.2024.2325846\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e\u0026Ccedil;ene E. What is the difference between a winning and a losing team: insights from Euroleague basketball. Int J Perf Anal Spor. 2018;18(1):55\u0026ndash;68. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/24748668.2018.1446234\u003c/span\u003e\u003cspan address=\"10.1080/24748668.2018.1446234\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCabarkapa D, Deane MA, Fry AC et al. Game statistics that discriminate winning and losing at the NBA level of basketball competition. Abi\u0026aacute;n-Vic\u0026eacute;n J, ed. \u003cem\u003ePLoS ONE\u003c/em\u003e. 2022;17(8): e0273427. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pone.0273427\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0273427\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMikołajec K, Banyś D, Żurowska-Cegielska J, Zawartka M, Gryko K. How to Win the Basketball Euroleague? Game Performance Determining Sports Results During 2003\u0026ndash;2016 Matches. J Hum Kinetics. 2021;77:287\u0026ndash;96. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.2478/hukin-2021-0050\u003c/span\u003e\u003cspan address=\"10.2478/hukin-2021-0050\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eConte D, Lukonaitienė I. Scoring Strategies Differentiating between Winning and Losing Teams during FIBA EuroBasket Women 2017. \u003cem\u003eSports\u003c/em\u003e. Published online. 2018. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/sports6020050\u003c/span\u003e\u003cspan address=\"10.3390/sports6020050\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTeramoto M, Cross CL, Rieger RH, Maak TG, Willick SE. Predictive Validity of National Basketball Association Draft Combine on Future Performance. J Strength Conditioning Res. 2018;32(2):396\u0026ndash;408. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1519/JSC.0000000000001798\u003c/span\u003e\u003cspan address=\"10.1519/JSC.0000000000001798\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSampaio J, Drinkwater EJ, Leite NM. Effects of season period, team quality, and playing time on basketball players\u0026rsquo; game-related statistics. Eur J Sport Sci. 2010;10(2):141\u0026ndash;9. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/17461390903311935\u003c/span\u003e\u003cspan address=\"10.1080/17461390903311935\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIb\u0026aacute;\u0026ntilde;ez SJ, Garc\u0026iacute;a J, Feu S, Lorenzo A, Sampaio J. Effects of consecutive basketball games on the game-related statistics that discriminate winner and losing teams. J Sport Sci Med. 2009;(8):458\u0026ndash;62.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eG\u0026oacute;mez M\u0026Aacute;, Lorenzo A, Sampaio J, Ib\u0026aacute;\u0026ntilde;ez SJ, Ortega E. Game-Related Statistics that Discriminated Winning and Losing Teams from the Spanish Men\u0026rsquo;s Professional Basketball Teams. Coll Antropol. 2008;32(2):315\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLorenzo A, G\u0026oacute;mez M\u0026Aacute;, Ortega E, Ib\u0026aacute;\u0026ntilde;ez SJ. Game related statistics which discriminate between winning and losing under-16 male basketball games. J Sport Sci Med. 2010;(9):664\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCsataljay G, O\u0026rsquo;Donoghue P, Hughes M, Dancs H. Performance indicators that distinguish winning and losing teams in basketball. Int J Perform Anal Sport. 2009;9(1):60\u0026ndash;6. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/24748668.2009.11868464\u003c/span\u003e\u003cspan address=\"10.1080/24748668.2009.11868464\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eG\u0026oacute;mez MA, Ib\u0026aacute;\u0026ntilde;ez SJ. The use of classification and regression tree when classifying winning and losing basketball teams. Kinesiology. 2017;49(1):47. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.26582/k.49.1.9\u003c/span\u003e\u003cspan address=\"10.26582/k.49.1.9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMarmarinos C, Apostolidis N, Kostopoulos N, Apostolidis A. Efficacy of the pick and roll offense in top level European basketball teams. J Hum Kinet. 2016;51(1):121\u0026ndash;9. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1515/hukin-2015-0176\u003c/span\u003e\u003cspan address=\"10.1515/hukin-2015-0176\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eConte D, Favero T, Niederhausen M, Capranica L, Tessitore A. Determinants of the effectiveness of fast break actions in elite and sub-elite Italian men\u0026rsquo;s basketball games. \u003cem\u003ebs\u003c/em\u003e. 2017; 2:177\u0026ndash;183. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.5114/biolsport.2017.65337\u003c/span\u003e\u003cspan address=\"10.5114/biolsport.2017.65337\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZukolo Z, Dizdar D, Selmanović A, Vidranski T. The role of finishing actions in the final result of the basketball match. Sport Sci. 2019;19:90\u0026ndash;5.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKoutsouridis C, Karamousalidis G, Galazoulas C. The efficacy of High Pick and Roll in relation to the defence\u0026rsquo;s reaction and its effect on the result of the game. Int J Perform Anal Sport. 2018;18(4):554\u0026ndash;67. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/24748668.2018.1501983\u003c/span\u003e\u003cspan address=\"10.1080/24748668.2018.1501983\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLorenzo Calvo J, Men\u0026eacute;ndez Garc\u0026iacute;a A, Navandar A. Analysis of mismatch after ball screens in Spanish professional basketball. Int J Perf Anal Spor. 2017;17(4):555\u0026ndash;62. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/24748668.2017.1367999\u003c/span\u003e\u003cspan address=\"10.1080/24748668.2017.1367999\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMatulaitis K, Bietkis T. Prediction of Offensive Possession Ends in Elite Basketball Teams. Int J Environ Res Public Health. 2021;18(3):1083. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/ijerph18031083\u003c/span\u003e\u003cspan address=\"10.3390/ijerph18031083\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFoteinakis P, Pavlidou S, Stavropoulos N. Analysis of the effectiveness of different play types in the end of game possessions of close EuroLeague matches. J Hum Sport Exerc. 2024;19(2):617\u0026ndash;30. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.55860/rj38hm02\u003c/span\u003e\u003cspan address=\"10.55860/rj38hm02\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBustamante-S\u0026aacute;nchez A, Gomez-Ruano MA, Clemente-Su\u0026aacute;rez VJ, Jim\u0026eacute;nez-S\u0026aacute;iz SL. Pre-shot combinations and game-related statistics discriminating between winners and losers depending on the game location during the NBA COVID-19 season. Front Physiol. 2022;13:949445. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fphys.2022.949445\u003c/span\u003e\u003cspan address=\"10.3389/fphys.2022.949445\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLeicht AS, G\u0026oacute;mez MA, Woods CT. Explaining Match Outcome during The Men\u0026rsquo;s Basketball Tournament at The Olympic Games. J Sports Sci Med. 2017;(16):468\u0026ndash;73.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSarlis V, Chatziilias V, Tjortjis C, Mandalidis D. A Data Science approach analysing the Impact of Injuries on Basketball Player and Team Performance. Inform Syst. 2021;99:101750. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.is.2021.101750\u003c/span\u003e\u003cspan address=\"10.1016/j.is.2021.101750\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSarlis V, Tjortjis C. Sports analytics \u0026mdash; Evaluation of basketball players and team performance. Inform Syst. 2020;93:101562. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.is.2020.101562\u003c/span\u003e\u003cspan address=\"10.1016/j.is.2020.101562\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTian C, De Silva V, Caine M, Swanson S. Use of Machine Learning to Automate the Identification of Basketball Strategies Using Whole Team Player Tracking Data. Appl Sci. 2019;10(1):24. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/app10010024\u003c/span\u003e\u003cspan address=\"10.3390/app10010024\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eQiu M, Zhang S, Yi Q, Zhou C, Zhang M. The influence of momentum on the game outcome while controlling for game types in basketball. Front Psychol. 2024;15:1412840. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fpsyg.2024.1412840\u003c/span\u003e\u003cspan address=\"10.3389/fpsyg.2024.1412840\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eThabtah F, Zhang L, Abdelhamid N. NBA Game Result Prediction Using Feature Analysis and Machine Learning. Ann Data Sci. 2019;6(1):103\u0026ndash;16. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s40745-018-00189-x\u003c/span\u003e\u003cspan address=\"10.1007/s40745-018-00189-x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCai W, Yu D, Wu Z, Du X, Zhou T. A hybrid ensemble learning framework for basketball outcomes prediction. Physica A. 2019;528:121461. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.physa.2019.121461\u003c/span\u003e\u003cspan address=\"10.1016/j.physa.2019.121461\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHarris AR, Roebber PJ. NBA team home advantage: Identifying key factors using an artificial neural network. PLoS ONE. 2019;14(7):e0220630. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pone.0220630\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0220630\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMigliorati M. Detecting drivers of basketball successful games: an exploratory study with machine learning algorithms. Electron J Appl Stat Anal. 2020;13(2):454\u0026ndash;73. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1285/I20705948V13N2P454\u003c/span\u003e\u003cspan address=\"10.1285/I20705948V13N2P454\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEkanayake IU, Meddage DPP, Rathnayake U. A novel approach to explain the black-box nature of machine learning in compressive strength predictions of concrete using Shapley additive explanations (SHAP). Case Stud Constr Mater. 2022;16:e01059. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.cscm. 2022.e01059\u003c/span\u003e\u003cspan address=\"10.1016/j.cscm. 2022.e01059\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOuyang Y, Li X, Zhou W et al. Integration of machine learning XGBoost and SHAP models for NBA game outcome prediction and quantitative analysis methodology. Ribeiro HV, ed. \u003cem\u003ePLoS ONE\u003c/em\u003e. 2024;19(7): e0307478. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pone.0307478\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0307478\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang X, Han B, Zhang S, Zhang L, Lorenzo Calvo A, Gomez M\u0026Aacute;. The Differences in the Performance Profiles Between Native and Foreign Players in the Chinese Basketball Association. Front Psychol. 2022;12. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fpsyg.2021.788498\u003c/span\u003e\u003cspan address=\"10.3389/fpsyg.2021.788498\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen R, Zhang M, Xu X. Modeling the influence of basketball players\u0026rsquo; offense roles on team performance. Front Psychol. 2023;14. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fpsyg.2023.1256796\u003c/span\u003e\u003cspan address=\"10.3389/fpsyg.2023.1256796\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eG\u0026oacute;mez M\u0026Aacute;, Medina R, Leicht AS, Zhang S, Vaquera A. The Performance Evolution of Match Play Styles in the Spanish Professional Basketball League. Appl Sci. 2020;10(20):7056. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/app10207056\u003c/span\u003e\u003cspan address=\"10.3390/app10207056\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang S, Lorenzo A, Woods CT, Leicht AS, G\u0026oacute;mez MA. Evolution of game-play characteristics within-season for the National Basketball Association. Int J Sports Sci Coaching. 2019;14(3):355\u0026ndash;62. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1177/1747954119847171\u003c/span\u003e\u003cspan address=\"10.1177/1747954119847171\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi Mzi, Lu Y, Chen Z, hao, Yang Q, Tao Yliu. Analysis of Key Performance Indicators and Major Winning Rules in Chinese Basketball Professional League. Published online Oct. 2023;10. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.21203/rs.3.rs-3407093/v1\u003c/span\u003e\u003cspan address=\"10.21203/rs.3.rs-3407093/v1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGomez MA, Gasperi L, Lupo C. Performance analysis of game dynamics during the 4th game quarter of NBA close games. Int J Perform Anal Sport. 2016;16(1):249\u0026ndash;63. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/24748668.2016.11868884\u003c/span\u003e\u003cspan address=\"10.1080/24748668.2016.11868884\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePapageorgiou G, Sarlis V, Tjortjis C. Evaluating the effectiveness of machine learning models for performance forecasting in basketball: a comparative study. Knowl Inf Syst. 2024;66(7):4333\u0026ndash;75. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s10115-024-02092-9\u003c/span\u003e\u003cspan address=\"10.1007/s10115-024-02092-9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePlakias S, Kokkotis C, Pantazis D, Tsatalas T. Comparative analysis of key performance indicators in Euroleague and national basketball leagues. J Phys Educ Sport. 2024;24(6):1360\u0026ndash;72. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.7752/jpes.2024.06154\u003c/span\u003e\u003cspan address=\"10.7752/jpes.2024.06154\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBustamante-S\u0026aacute;nchez \u0026Aacute;, G\u0026oacute;mez MA, Jim\u0026eacute;nez-Saiz SL. Game location effect in the NBA: A comparative analysis of playing at home, away and in a neutral court during the COVID-19 season. Int J Perf Anal Spor. 2022;22(3):370\u0026ndash;81. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/24748668.2022.2062178\u003c/span\u003e\u003cspan address=\"10.1080/24748668.2022.2062178\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKrzyszkowski J, Kipp K. Prediction of throwing distance in the men\u0026rsquo;s and women\u0026rsquo;s javelin final at the 2017 IAAF world championships. J Sports Sci. 2021;39(9):1055\u0026ndash;9. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/02640414.2020.1856462\u003c/span\u003e\u003cspan address=\"10.1080/02640414.2020.1856462\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKolodziej M, Groll A, Nolte K, et al. Predictive modeling of lower extremity injury risk in male elite youth soccer players using least absolute shrinkage and selection operator regression. Scandinavian Med Sci Sports. 2023;33(6):1021\u0026ndash;33. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1111/sms.14322\u003c/span\u003e\u003cspan address=\"10.1111/sms.14322\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eScanlan A, Humphries B, Tucker PS, Dalbo V. The influence of physical and cognitive factors on reactive agility performance in men basketball players. J Sports Sci. 2014;32(4):367\u0026ndash;74. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/02640414.2013.825730\u003c/span\u003e\u003cspan address=\"10.1080/02640414.2013.825730\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi Z. Extracting spatial effects from machine learning model using local interpretation method: An example of SHAP and XGBoost. Comput Environ Urban Syst. 2022;96:101845. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.compenvurbsys.2022.101845\u003c/span\u003e\u003cspan address=\"10.1016/j.compenvurbsys.2022.101845\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStavropoulos N, Kolias P, Papadopoulou A, et al. Game related predictors discriminating between winning and losing teams in preliminary, second and final round of basketball world cup 2019. Int J Perform Anal Sport. 2021;21(3):383\u0026ndash;95. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/24748668.2021.1901437\u003c/span\u003e\u003cspan address=\"10.1080/24748668.2021.1901437\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang S, Lorenzo A, Zhou C, Cui Y, Gon\u0026ccedil;alves B, Angel G\u0026oacute;mez M. Performance profiles and opposition interaction during game-play in elite basketball: evidences from National Basketball Association. Int J Perform Anal Sport. 2019;19(1):28\u0026ndash;48. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/24748668.2018.1555738\u003c/span\u003e\u003cspan address=\"10.1080/24748668.2018.1555738\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKubatko J, Oliver D, Pelton K, Rosenbaum DT. A Starting Point for Analyzing Basketball Statistics. \u003cem\u003eJournal of Quantitative Analysis in Sports\u003c/em\u003e. 2007;3(3): Article 1. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.2202/1559-0410.1070\u003c/span\u003e\u003cspan address=\"10.2202/1559-0410.1070\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCs\u0026aacute;taljay G, James N, Hughes M, Dancs H. Analysis of influencing factors behind offensive rebounding performance in elite basketball. Int J Sports Sci Coa. 2017;12(6):774\u0026ndash;81. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1177/1747954117738900\u003c/span\u003e\u003cspan address=\"10.1177/1747954117738900\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMalarranha J, Figueira B, Leite N, et al. Dynamic Modeling of Performance in Basketball. Int J Perform Anal Sport. 2013;13(2):377\u0026ndash;87. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/24748668.2013.11868655\u003c/span\u003e\u003cspan address=\"10.1080/24748668.2013.11868655\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSampaio J, Lago C, Drinkwater EJ. Explanations for the United States of America\u0026rsquo;s dominance in basketball at the Beijing Olympic Games (2008). \u003cem\u003eJournal of Sports Sciences\u003c/em\u003e. 2010;28(2):147\u0026ndash;152. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/02640410903380486\u003c/span\u003e\u003cspan address=\"10.1080/02640410903380486\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRuano M, Avugos S, Ortega E, Bar-Eli M. Adverse effects of technical fouls in elite basketball performance. Biology Sport. 2019;36:155\u0026ndash;60. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.5114/biolsport.2019.83505\u003c/span\u003e\u003cspan address=\"10.5114/biolsport.2019.83505\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSampaio J, Leser R, Baca A, et al. Defensive pressure affects basketball technical actions but not the time-motion variables. J Sport Health Sci. 2016;5:375\u0026ndash;80. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.jshs.2015.01.011\u003c/span\u003e\u003cspan address=\"10.1016/j.jshs.2015.01.011\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eG\u0026oacute;mez M\u0026aacute;ngel, Lorenzo A, Jim\u0026eacute;nez S, Navarro RM, Sampaio J. Examining Choking in Basketball: Effects of Game Outcome and Situational Variables during Last 5 Minutes and Overtimes. Percept Mot Skills. 2015;120(1):111\u0026ndash;24. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.2466/25.29.PMS.120v11x0\u003c/span\u003e\u003cspan address=\"10.2466/25.29.PMS.120v11x0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eG\u0026oacute;mez MA, Lorenzo A, Iba\u0026ntilde;ez SJ, Sampaio J. Ball possession effectiveness in men\u0026rsquo;s and women\u0026rsquo;s elite basketball according to situational variables in different game periods. J Sport Sci. 2013;31(14):1578\u0026ndash;87. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/02640414.2013.792942\u003c/span\u003e\u003cspan address=\"10.1080/02640414.2013.792942\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStavropoulos N, Papadopoulou A, Kolias P. Evaluating the Efficiency of Off-Ball Screens in Elite Basketball Teams via Second-Order Markov Modelling. Mathematics. 2021;9(16):1991. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/math9161991\u003c/span\u003e\u003cspan address=\"10.3390/math9161991\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFoteinakis P, Pavlidou S. Positional Differences in the Efficacy of Critical End-of-Game Possessions in EuroLeague Basketball. Sport Mont. 2024;22(2):25\u0026ndash;31. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.26773/smj.240704\u003c/span\u003e\u003cspan address=\"10.26773/smj.240704\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-sports-science-medicine-and-rehabilitation","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"ssmr","sideBox":"Learn more about [BMC Sports Science, Medicine and Rehabilitation](http://bmcsportsscimedrehabil.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/ssmr/default.aspx","title":"BMC Sports Science, Medicine and Rehabilitation","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"basketball, key performance indicators, CBA, play-type, machine learning classifiers","lastPublishedDoi":"10.21203/rs.3.rs-6230128/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6230128/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThis study investigated the technical and tactical variables related to success in professional basketball, utilizing data from 3,249 Chinese Basketball Association (CBA) games during the 2018\u0026ndash;2022 seasons. Key performance indicators (KPIs) were identified through lasso and stepwise regression, highlighting assists, fouls drawn, offensive rebound percentage, steals, blocks, defensive rebounds, and successful screens-off and pick-and-roll handler\u0026rsquo;s shots as positively correlated with winning outcomes, while turnovers, missed shots, and fouls demonstrated negative associations. To predict game outcomes, ten machine learning (ML) models were developed, with Artificial Neural Networks (AUC\u0026thinsp;=\u0026thinsp;0.914), Support Vector Machines (AUC\u0026thinsp;=\u0026thinsp;0.913), Logistic Regression (AUC\u0026thinsp;=\u0026thinsp;0.899), and XGBoost (AUC\u0026thinsp;=\u0026thinsp;0.897) achieving the higher performance. The SHapley Additive exPlanations (SHAP) algorithm further enhanced interpretability, quantifying defensive rebounds, offensive rebound percentage, and missed two-point field attempts consistently as the most influential KPIs. This study provides a comprehensive analysis of game-related and play-type statistics, offering further insights into how they work together for optimizing the multifaceted dynamics of basketball performance. The findings prioritize possession retention through rebounding and effective two-point shot selection in predicting the desired game outcome, presenting how ML-driven KPI identification inform coaching game strategies and refine opponent-specific preparations.\u003c/p\u003e","manuscriptTitle":"Comparison of machine learning classifiers for identifying key performance indicators in the Chinese Basketball Association","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-03-31 13:04:03","doi":"10.21203/rs.3.rs-6230128/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-11-25T10:07:34+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-11-24T21:38:35+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"110717747795462952837783961941261637013","date":"2025-07-28T10:35:41+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-04-15T15:28:50+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"296392685306282731105478645217548141883","date":"2025-03-24T11:59:58+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"79018846355643888459258734760086622552","date":"2025-03-22T16:42:11+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-03-22T10:54:40+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-03-21T06:15:14+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-03-21T02:44:07+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Sports Science, Medicine and Rehabilitation","date":"2025-03-21T02:43:00+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-sports-science-medicine-and-rehabilitation","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"ssmr","sideBox":"Learn more about [BMC Sports Science, Medicine and Rehabilitation](http://bmcsportsscimedrehabil.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/ssmr/default.aspx","title":"BMC Sports Science, Medicine and Rehabilitation","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"d7a67430-6f27-4eae-af65-bb000e226b10","owner":[],"postedDate":"March 31st, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2026-03-16T16:05:53+00:00","versionOfRecord":{"articleIdentity":"rs-6230128","link":"https://doi.org/10.1186/s13102-026-01620-0","journal":{"identity":"bmc-sports-science-medicine-and-rehabilitation","isVorOnly":false,"title":"BMC Sports Science, Medicine and Rehabilitation"},"publishedOn":"2026-03-10 15:58:09","publishedOnDateReadable":"March 10th, 2026"},"versionCreatedAt":"2025-03-31 13:04:03","video":"","vorDoi":"10.1186/s13102-026-01620-0","vorDoiUrl":"https://doi.org/10.1186/s13102-026-01620-0","workflowStages":[]},"version":"v1","identity":"rs-6230128","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6230128","identity":"rs-6230128","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}