The 2016-2017 NBA schedule was released yesterday
and the reactions are pouring in
across the sports media landscape. For good reason, the focus is on the marquee matchups, the Christmas Day lineup, and the epic grudge match between Oklahoma City and Golden State in February. But the release of the schedule also means a new test case for the ideas we've been exploring here over the past few weeks. In case you missed the past installments, you can find them here
, and here
. But before we jump into the new season, as promised, we will expand the scope of our analysis beyond the Atlanta Hawks to the league as a whole – with a focus on rest days.
Over the past few weeks, we've hypothesized that there are structural advantages built into the schedule by way of days of rest in between games. One of these advantages comes in the form of more rest than the opponent. On average, we would expect the more well-rested team to perform better. Up to this point we've only looked at the Atlanta Hawks and their past two seasons. In the table following table, we can see the performance of every team over the past two seasons with a rest advantage.
To make it easier for comparison, I've presented the advantage games played as a percentage of total games for each team. The next column is the win percentage in these advantage games, followed by the win percentage in games where the team and opponent had equal rest. Teams play the majority of their games against opponents in an equal rest setting, so it provides a benchmark for the advantage of playing with more rest. Finally, the last column estimates how many wins the additional rest may have been worth to each team in a single season. By taking the difference between the win percentage columns – gathered from the past two seasons – and multiplying by the average number of games with a rest advantage we get the number of wins above equal rest.
With a much larger sample we can begin to understand the true effect of additional days of rest on the outcomes of games. As you can see, on the average, over the past two seasons there is both a noticeable difference in team performance and impact on wins due to rest advantage. The win percentage difference (4.9%) would be worth just over four wins over 82 games. However, the schedule has not afforded 82 games of rest advantage, only 21 on average (25.7% of 82). Therefore, the impact of structural rest advantage has been worth just 1.03 wins.
To bring it back to the Atlanta Hawks for a moment, we now have some benchmarks to understand better how the schedule weighed on their results in the context of the entire league. We can now see that the Hawks got an unlucky draw, with the second-fewest percentage of games played with a rest advantage. Despite the unfair schedule, Atlanta still managed an above-average win difference, a testament to how good the Hawks have been over the past two seasons. This is only half the picture however, because they also played games with less rest than their opponents – rest disadvantage.
Using the same methodology as above, the table below uses the past two season to estimate the impact of being at a rest disadvantage has on performance. If teams perform better when more well-rested, it should follow that teams perfom worse when less rested than opponents. Taking the two tables together we can get a feel for how rest advantage has played out on wins and losses in the NBA the past two seasons.
On the average, the results for rest disadvantage are the opposite of those with a rest advantage – minus four wins over 82 games and an average real impact of -1.03 wins. But as we've learned over the past several weeks, the schedule is not fair so the impact is not the exact opposite for each team. Take Atlanta for example – their below average percentage of games with a rest advantage was not met with an above average percentage with a rest disadnvantage. As such, the Hawks managed to see only a -0.37 win difference due to disadvantageous rest scenarios. Together, the Hawks saw an overall impact of 0.74 wins on their record looking through this lens.
The rest advantage view of the schedule is powerful and simple, but it does not consider the number of days of rest. Additionally, with the majority of games being played with equal rest, we need more information to uncover the extent of the impact of the schedule, particularly rest. Therefore, we reintroduce the number of days off, this time aggregated over the entire league over the past two seasons. Again we will use the methodology from above to compare the days off categories. The only difference is that the comparison win percentage to estimate the win difference is the average win percentage of .500.
As we would expect, the lowest win percentage has come with zero days rest – also known as a back to back. The win percentages continue to rise as expected with more rest. Interestingly, the win percentage dips with three or more days rest, though I suspect that may be a product of a smallish sample size. Although, there may be something to too much rest that breaks the rhythm of the season. Either way, more research is necessary to close the book on that.
The efforts of the NBA office to reduce the back-to-backs league-wide has neutered the win/loss impact of zero days rest. The win percentage difference would be worth -3.6 wins over a full season. Even still, playing on zero days rest has the largest impact. Considering the two highest probability rest day groups, the likeliest rest advantage situations are between a team with zero days rest and another with one day of rest. The combined win difference is 1.2 wins, slightly higher than the average 1.03 win difference we saw earlier – these games are dragging the average win difference for rest advantage higher. As I mentioned, not every rest advantage situation is created equal.
After several weeks of looking at the schedule effects, these numbers may seem insignificant and all this effort and reading and writing may not have been worth it. However, those thoughts would be misguided. Of course we are working on the margins, no schedule will save the Brooklyn Nets next season. But, the entire point of analytics and thinking deeper about the game is to identify marginal advantages in order to exploit them in a big way – the entire concept of Moneyball
. Consider the Eastern Conference last season. The three through six seeds had the exact same record. A one win difference that we saw with the rest advantage is huge in that context, the difference between game one of the playoffs at home or on the road. I am very excited by what we've uncovered this week, legitimizing the concept of rest advantage. Next week, we will pivot over to win transfers on a larger scale to see put what we saw with Atlanta into context and see if there are signals worth pursuing when we finally analyze and predict the effects of the new NBA schedule on this upcoming season.