超級籃球聯賽之進階攻守數據研究

Abstract

運動數據分析可謂近三十年來重要趨勢，各類籃球進階攻守數據模型以數學統計方法分析比賽結果及運動表現，除了比起傳統的基本攻守統計 (Box Score) 有更好的預測力與解釋力，並能解釋更多場內外之現象，進而提供教練團、球隊管理階層及場外有關人士更多有用資訊。目的：應用各種進階數據於超級籃球聯賽分析，探討適合超級籃球賽分析者。方法：本研究收集各種進階攻守數據、球員表現進階數據模型及比賽結果預測模式，計算超級籃球聯賽之分析結果，探討其適用性與解釋力。結果：一、各進階攻守數據能夠有效解釋各項基本攻守統計數據背後的效率表現。二、各進階數據模型能夠分析超級籃球聯賽球員整體表現，其中勝場貢獻值最能有效預測超級籃球聯賽個人獎項。三、各比賽結果預測模是皆能解釋90% 以上的勝負結果，其中鐘型曲線最為優異。結論：各種進階攻守數據模型能夠有效分析超級籃球聯賽球隊、球員表現與預測比賽結果，得從中再加以探討各種影響因素。Basketball analytics is an increasing trend in the past thirty decades. Advanced statistics models show better predictive and explanatory power than the traditional box score views. The purpose of this study is to analyze the productivity and efficiency of teams and players in the Taiwanese Super Basketball League (SBL) by using various basketball analytic models. The result shows that each analytic model can be used in analyzing SBL after appropriate modifications and adjustments for its coefficients or calculation methods. The average possessions per game in the 13th SBL were about 78, which is the foundation of most analytic models used in our study. The Win Shares model shows better explanatory power of wins. Moreover, it is relatively accurate in predicting individual awards in SBL. Besides, the Bell Curve method has the optimalaccuracy in winning predictions. In conclusion, we can use those analytic models to measure the factors influencing productivity and efficiency of teams and players in SBL.