Although there have been plenty of articles debating these two points, even many that apply some sort of statistical analysis, we figured it would be interesting to examine these claims using betting data – a relatively unique but insightful dataset. So, without further ado, let’s play Mythbusters with two popular claims about NBA playoff basketball.
Claim No. 1: Home-court advantage matters more in the playoffs
There is no doubt that home-court advantage matters a lot in the NBA, even relative to other leagues. For the dataset we have, which contains NBA games going back to the 1989-1990 season, during the regular season home teams have won at 60.29 percent rate, although that has been gradually changing over time, as evidenced in the below graphic.
That number is even greater for the postseason – with about 64.8 percent of home teams winning during the playoffs. However, that number is likely to be biased, as more often than not the higher seed – and presumably the stronger team – plays at home. So how can we see if home-court advantage is truly stronger in the postseason?
Well, if it is stronger, there are two possibilities: either the line does not account for the difference, in which case there is some arbitrage scenario, or the line does account for it and there is no arbitrage scenario.
Looking at the first possibility, we can see that home teams have a 50.55 percent cover rate against the spread in the playoffs. So only marginally higher than fifty percent, and nowhere close enough to be significant either practically or statistically. For reference, the 95 percent confidence interval is 48.28 percent – 52.8 percent – so it is more likely than not that there is no advantage to betting home teams in the playoffs.
So now, let’s examine the second possibility that the lines somehow account for this difference in home-court advantage. To do this, we can just compare the lines in the playoffs to the lines that were present during the regular season. Because every team plays every other team both home and away, those lines are guaranteed to exist.
Is this perfect? No. Perhaps a team was missing a key player when they played during the regular season. This method does not account for that. However, with a large enough sample size, more than 1900 observations, that variance should hopefully not matter much.
Doing this, we can see that on average, home teams were favored by 0.2 more points in the postseason compared to when they played in the regular season. That suggests a marginal but significant difference: the p-value associated with that difference is 0.004, and the 95 percent confidence interval is 0.35 to 0.06. However, the change in home-court advantage, as measured this way, despite being significant, is very small.
In conclusion: there does seem to be some evidence for a larger home-court advantage in the playoffs, although it is not that significant.
Claim No. 2: Scoring is depressed in the playoffs
Again, the naive thing to do here would just be to compare average points per game in the regular season to points per game in the playoffs. This has been done many times before, and while the average points per game (both teams) is 197, that number drops to 190 in the playoffs. But that doesn’t mean much – perhaps the postseason is full of teams that play above average defense (“Defense wins championships?”).
So in order to test this, we can use the same methodology as above, namely, we can look at whether there is a systematic advantage to betting the Over (or the Under) in the postseason, and whether the total lines reflect a difference in expected points per game.
Tackling the first part, we can see pretty immediately that there is no real advantage to one side of the total bet. Overall, 48.3 percent of games go Under the total in the playoffs, which is close to (but not quite) significant. The 95 percent confidence interval is 46 percent – 50.6 percent, and the p-value is 0.15. Furthermore, the discrepancy is so small (only a percent and change) that even if it were “statistically significant” it might not be practical to always bet the Under.
So do lines reflect lower expected scoring? Or is there just no change in scoring in the postseason? Using the same methodology as we used for home-court advantage, we can see that lines are significantly lower in the postseason.
On average, total lines are 3.64 points lower in the postseason than they were in the regular season. This is incredibly significant, and does seem to suggest that scoring is lower in the playoffs – although, because the lines do take that into account.
There are several reasons we can imagine why this might be: defenses try harder, better defensive teams are in the postseason, and probably many more I am not thinking of. It is tough to tease out exactly why scoring is lower, all we know for sure is that is significantly lower in the postseason.
So, in conclusion, both of these claims about the postseason appear to be true, especially the one regarding teams scoring at lower rates. Unfortunately for NBA bettors, the lines appear to already factor in all this information, so there isn’t a huge edge to be gained.
However, one little tidbit we did discover when looking at the data is that when total lines in the postseason are greater than they were in the regular, it is more likely that the game will go Under the total.
For example, only about 25 percent of postseason games had a higher total than in the regular season, but those games went Under the total 46.8 percent of the time. Although that is not quite statistically significant, there may be enough of an edge there to gain some advantage in the long run.
Harrison Chase is the Co-President of the Harvard Sports Analysis Collective, a student-run organization at Harvard College dedicated to the quantitative analysis of sports strategy and management.
Tagged with: feature