Investigation of the determinants of match winning in the Turkish Basketball League using fourier VAR and causality methods

Analyzing the determinants of match-winning in basketball is critical for teams' training and tactical decisions. This thesis analyzes the determinants of match-winning using Fourier-based econometric methods. Twelve seasons of data from Anadolu Efes, Fenerbahçe, Galatasaray, Beşiktaş, and Karşıyaka, selected from the TBF, were used to analyze these determinants. The predictors of match-winning for these teams were shot count and percentages, defensive and offensive determinants, and ball-handling determinants. The predictor variables were analyzed separately for each team. Stationarity of the variables was determined using Fourier-based unit root tests. After determining stationarity, the variables were analyzed using the Fourier VAR model, and a Fourier-based causality test was applied. Fourier unit root test results indicate that all variables exhibited stationarity at the level across teams. Different Fourier unit root tests, which revealed smooth transitions and structural breaks, indicate the necessity of Fourier terms for some variables across teams. The variables appear to have different short-term effects and responses across teams. Point difference, representing winning, exhibits a causal relationship across all teams over a four-week period. For Fenerbahçe, the point difference variable representing match winning has a short-term causal relationship with three-point shooting percentage. The contributions of the determinants of match winning appear to vary for each team. It is believed that point difference representing match winning may be determined not only by the variables included in the model but also by the teams' different structural characteristics and their changing playing strategies over time. In this context, policy recommendations for technical teams and analysis units should be addressed on a team-by-team basis.