ROCKETS VS THUNDER
In anticipation for game 7 of an amazing series, let’s take a closer look at how the Thunder and Rockets performed during the regular season and see if we can gain some insights.
What we want to do is compare how a team performs in a certain statistical category in wins, and compare that to how a team performs in losses. If there is no statistically significant difference between the two groups then we can say that the distributions are similar and that the team generally does the same thing in both wins and losses.
To test for statistical significance, we can use the Kruskal test to see whether the distributions of two or more independent samples are equal or not. Let’s state our hypothesis as: the distributions are equal.
The test will give us a p-value, and using a confidence level of 95%, if the p-value is less than 0.05 (meaning the differences in the distributions is statistically significant) then we should reject our hypothesis. A p-value less than 0.05 tells us that the distributions of a category when grouped by wins and losses are different, which may tell us something about what a team does during winning and losing efforts.
We can perform this test for each category grouped by W and L and rank the ones with the lowest p value, which will give us an idea of what categories are important to a team’s success.
HOU p-values
Below we can see a list of categories with a p-value < 0.05, aka a list of categories where the Rockets performed differently for Wins and Losses. To no one’s surprise, this list is heavily populated with a variety of 3PT shooting stats. Other stats to keep note of are scoring in the paint (AtRimAccuracy, AtRimPctBlocked) and Rebounding.
Here’s a few of the distributions split between W and L to visualize some of the differences.
Let’s do the same and look at how HOU’s opponents do during HOU Wins and Losses.
HOU OPPONENT p-values
We can see that the 3 is also important for HOU’s opponents, as well as rebounding. It also appears that HOU opponents record more blocks (Blocked2s, BlockedAtRim, Blocks) in winning efforts.
Again, here are some examples of the distributions.
Summary
The Houston Rockets should look to make a lot of 3’s and finish at the rim at a high clip.
Rocket Opponents need to also shoot 3’s well, rebound, and block/contest some layups and shots at the rim.
OKC p-values
A lot of categories show up in this list, suggesting that the OKC are a little more inconsistent with their levels of play. Keys for OKC appear to be scoring at the rim (field goals at the rim, as well as assists at the rim) and Rebounding.
OKC OPPONENT p-values
Again we see rebounding playing an important factor here. We can also see that OKC losses appear to be correlated to an opponent shooting well from 3 and not having their 2pt field goals attempts blocked.
Summary
The OKC Thunder will want to finish at the rim, rebound, and protect the paint to give them a better chance at a win.
In general, OKC opponents need to rebound, and shoot 3’s well, as well as not have their 2pt field goal attempts blocked.
REBOUNDING AND 3'S
Based on this analysis, it really does make sense that this series would be so tightly contested. The Houston Rockets are great at shooting 3’s, which is what opponents that win against the Thunder appear to do well. On the other hand, the Thunder lead the playoffs in rebounds, which is what HOU opponents generally do well when they win.
With HOU holding an advantage in 3’s, and OKC an advantage in rebounding, it’ll likely come down to which team can close the gap in the other category.
Data from PBPstats.com
Analysis done using Python and Jupyter Notebooks
