Written by: Justin Lee
In Correspondence to Sam Ventura’s class: 36-144 (Winning with Statistics in Sports)
Despite the grueling 82 game schedule of a NBA team, with its long road trips and back-to-back games, the amount of rest a player gets and how that affects their team’s performance is often overlooked. Players rarely blame fatigue, travel, and back-to-backs as reasons for their losses and poor performance. For example, when the Golden State Warriors’ 24 game win streak was busted by the Milwaukee Bucks in 2016, how much did the fact that the Warriors just came off a double overtime victory play a role in that loss? More generally, how much of a difference does rest make and what amount is ideal for teams?
To measure the impact of rest, I used statistics from teamrankings.com and categorized rest into 0 days, 1 day, 2-3 days, and 4+ days. I looked at how the various amounts of rest affects win percentage of teams as well as the margin of victory (MOV). For both response variables, I recorded the win percentage and MOV of every NBA team since 2003, and took the average of those values. Since so many games were played in that period of time, we could assume that the variability of difficulty of teams played was neutralized. By taking the average across all teams, we yielded a general NBA win percentage with the corresponding amounts of rest without biased caused by differences in team talent, conditioning, strategy, health, etc.
Analysis of Win Percentage
Graph 1: Mean Win Percentage vs. Amount of Rest
Based on the graph, it seems that the difference between rest days of 1, 2-3, and 4+ is essentially non-existent. However, for 0 days of rest, the win percentage is around 7-8 percent lower. To test whether this difference was significant, I ran a One-Way ANOVA Test. To confirm that there was no significant difference in mean win percentage amongst 1 day, 2-3 days, and 4+ days of rest, I first did a One-Way ANOVA Test for just those three categorical variables. The p-value of this test was 0.898 which is greater than a significance level of 0.05, showing there is not enough evidence to conclude that any of the means are different. In other words, it shows that there is no significant difference in win percentage between 1, 2-3, and 4+ days of rest. For the next test, I introduced the fourth category of “0 days” of rest. Here I got a p-value of 0.003 which is smaller than 0.05, showing that there is significant evidence to conclude that not all means of win percentage are the same. The fact that introducing the “0 days” category changed the p-value so significantly shows that the win percentage of teams after 0 days of rest, is significantly lower than the other amounts of rest.
Analysis of Margin of Victory
Graph 2: Mean Margin of Victory vs. Amount of Rest
For the analysis of margin of victory, I followed the same process as i did for mean win percentage. I performed a One-Way ANOVA test on the categories 1 day, 2-3 days, and 4+ days of rest. The p-value was 0.96, which is clearly higher than 0.05 significance level, showing that there was not enough evidence to conclude that any of the mean MOV of those categories are different. However, when adding the fourth categorical variable of 0 days of rest, the p-value was 0.004. This is less than 0.05, showing there is enough evidence to conclude that the mean margin of victory for teams with no days of rests, is significantly lower than those of more rested teams.
To sum up, the margin of victory and the win percentage of teams coming off of 0 days of rest, was significantly lower than teams with 1, 2-3, and 4+ days of rest. This shows that having at least 1 day of rest is significantly beneficial to NBA players and teams, but having more than 1 day of rest is not necessarily more helpful than having just 1 day.
In addition to the fact that players are simply tired, a cause for these findings may be that on back to backs, coaches don’t necessarily sit their star players entirely but give them reduced minutes. The reduced playing time of these franchise players reduces the team’s chances of winning, which may also contribute to these results.
Using these findings, those who schedule the NBA season should be aware of how many back to back games each team plays, and attempt to reduce the amount of back to backs in a given season. So far, schedulers have been doing this. The average number of back to back games per team in the 2016-2017 season was 16.3 with a minimum of 13 (Oklahoma City) and a maximum of 19 (Atlanta). In the 2015-2016 season the average was 17.8 and the season before it was 19.3. Evidently, there has been improvement. If possible, this deviation should be reduced even further because those with more back to backs have a proven disadvantage.
Additionally, in order to beat the fatigue effect of back to backs, teams could reconsider how they train and condition. This brings up the hot topic in the NBA of resting healthy players. What teams have been doing recently is resting their star players, despite being healthy. This is understandable on a fatigue and health standpoint, especially for back to backs. However, this is unfair to fans who paid to watch these players. The NBA is indeed a market, and league leaders are upset by coaches resting their players randomly. These findings can provide insight on how to handle that situation in the future. Knowing this data, will teams rest healthy players more often? How should the NBA react and will they create a policy to prevent this from happening?
Essentially, my findings show that back to backs take a significant toll on NBA players, hurting their performance. The NBA and individual teams should continue to do research on rest to make new strides in the process of scheduling the season and in solving the controversy over resting healthy players.