Fig 1.
A visual summary of changes over time in player rotation (Entropy.MIN) and number of possessions per game (F.NP). The points indicate seasonal averages. The lines and shaded areas are loess smoothed individual (per game) observations with standard errors. Results are broken down by stage (regular season vs playoffs).
Table 1.
A numerical comparison of the Euroleague and the NBA (regular season and playoffs combined).
The first two columns are the estimated means (μ), followed by estimated difference in means (Δμ), and posterior probability that the difference is positive (). The fifth and sixth column are the estimates slopes (s), followed by estimated difference in slope (Δs), and posterior probability that the difference is positive (
). Posterior probabilities under 0.01 or over 0.99 are marked with *. The second line for each variable contains the standard errors for the estimates. Note that the posterior distributions of μ and s are approximately normal so ± 2 standard errors is approximately the 95% Bayesian posterior confidence interval. Note that all estimates except for F.NP and Entropy.MIN are multiplied by 100 to simplify interpretation.
Fig 2.
A visual summary of changes over time in free-throw (FT), two-point (P2), and three-point (P3) shooting percentages (pct) and shots attempted (PaceAdj.Tot). The points indicate seasonal averages. The lines and shaded areas are loess smoothed individual (per game) observations with standard errors. Results are broken down by stage (regular season vs playoffs).
Fig 3.
A visual summary of changes over time in the number of assists (PaceAdj.Tot.AST), blocks (PaceAdj.Tot.BLK), steals (PaceAdj.Tot.STL), and turnovers (PaceAdj.Tot.TOV). The points indicate seasonal averages. The lines and shaded areas are loess smoothed individual (per game) observations with standard errors. Results are broken down by stage (regular season vs playoffs).
Fig 4.
A visual summary of changes over time in the number of free-throws attempted per foul committed (F.FTperF) and the total number of fouls committed (PaceAdj.Tot.FCM). The points indicate seasonal averages. The lines and shaded areas are loess smoothed individual (per game) observations with standard errors. Results are broken down by stage (regular season vs playoffs).
Fig 5.
A visual summary of changes over time in the defensive (F.DRB) and the offensive (F.ORB) rebounding rates and the total number of rebounds (PaceAdj.Tot.TRB). The points indicate seasonal averages. The lines and shaded areas are loess smoothed individual (per game) observations with standard errors. Results are broken down by stage (regular season vs playoffs).
Table 2.
A numerical comparison of the Euroleague regular season and Euroleague playoffs.
The first two columns are the estimated means (μ), followed by estimated difference in means (Δμ), and posterior probability that the difference is positive (). The fifth and sixth column are the estimates slopes (s), followed by estimated difference in slope (Δs), and posterior probability that the difference is positive (
). Posterior probabilities under 0.01 or over 0.99 are marked with *. The second line for each variable contains the standard errors for the estimates. Note that the posterior distributions of μ and s are approximately normal so ± 2 standard errors is approximately the 95% Bayesian posterior confidence interval. Note that all estimates except for F.NP and Entropy.MIN are multiplied by 100 to simplify interpretation.
Table 3.
A numerical comparison of the NBA regular season and NBA playoffs.
The first two columns are the estimated means (μ), followed by estimated difference in means (Δμ), and posterior probability that the difference is positive (). The fifth and sixth column are the estimates slopes (s), followed by estimated difference in slope (Δs), and posterior probability that the difference is positive (
). Posterior probabilities under 0.01 or over 0.99 are marked with *. The second line for each variable contains the standard errors for the estimates. Note that the posterior distributions of μ and s are approximately normal so ± 2 standard errors is approximately the 95% Bayesian posterior confidence interval. Note that all estimates except for F.NP and Entropy.MIN are multiplied by 100 to simplify interpretation.
Fig 6.
A visual summary of the estimated adjusted R2 values for the linear prediction model of team success. The dashed lines represent the estimates across all seasons.