How to Win at Basketball, or 6 Habits of Winning Basketball Teams
I looked at NBA stats for every team from 1979 to 2009 and looked at how each stat was most correlated with winning and with losing. This gives us a measure of which stats were most important for winning and most important for losing. Obviously, you want to do more of the things winning teams do and less of the things the losers do.
The data came from http://www.databasebasketball.com/stats_download.htm and the correlation was done using corrcoef in MATLAB. Before doing the correlation I sorted the stats by total wins, though I don’t think this makes a difference.
First let’s go over the top 6 things losing teams do, in order from least to more correlated to losing.
The stat is followed by its correlation to losing.
6) Steals allowed; 0.3498
5) Assists allowed; 0.3671
4) Points allowed; 0.3800
3) Total rebounds allowed; 0.3866;
2) Defensive rebounds allowed; 0.427
1) Blocks allowed; 0.5011
These correlations are relatively strong. The most interesting about this by itself is that three most important are not related to points directly and that turnovers did not make the top 6, though steals and blocks typically lead to turnovers.
The top 6 stats associated with winning somewhat reflect the ones related to losing, which makes sense since you would figure winning teams do the opposite of losing teams. However there are a couple of differences
6) Free throws made; 0.290
5) Field goals made; 0.290
4) Rebounds made; 0.3389
3) Points made; 0.3594
2) Assists; 0.388
1) Defensive Rebounds; 0.4447
Since some of these are highly correlated with each other, here are some other interesting stats that were mixed in.
Blocking shots and steals both just missed the top 6 for winners, with a correlations just less than three throws made.
Three pointers made and 3 pointers allowed had correlations close to 0 for winning.
Turnovers allowed were significantly correlated with losing, with correlation at about 0.33 .
Points scored was not correlated with losing; the correlation was close to 0.
So how do we interpret these results?
Well this sort of analysis is pretty boring unless it gives counter intuitive results. So let’s review the counter-intuitive stuff. Points scored was not related to losing! This means that for the bad teams, the problem is more about allowing points (IE bad defense) than it is about bad offense. The correlations with losing support this as points allowed and assist allowed are in the top 6. Still, it seems like inefficient offense is a big culprit, since getting blocked was the most significant factor. When you look at all the things correlated with losing, it seems like the big problem is ball control and defense. Losing teams don’t get rebounds, get blocked, and get the ball stolen and allow points.
On the winning side, we also see that controlling the ball is very important for winning. Winning teams get rebounds, make assists , block shots, and get steals.
Some miscellaneous observations
Field goals and free throws made were highly correlated but 3 pointers were not. This points to the fact that winning teams score efficiently and don’t rely on shooting threes.
There is an age old question in basketball as to whether or not offense is more important than defense. This analysis gives the edge to defense, though really it would be more correct to say that ball control is the most important, as well as offensive efficiency.