A Deeper Look Into NBA Per Possession Statistics
Recently Haralabos Voulgaris had a mini tweet storm about pace in the NBA and how we might not be using the smartest evaluation models for how fast an NBA team plays.
For those unaware, here is a little bit about how we currently calculate Pace courtesy of captaincalculator.com:
Pace Factor (Pace) is calculated as:
For example, in a game, a team has had 82 possessions and their opponents 77. The team’s players have played for a total of 240 minutes, and the match was 48 minutes long.
Therefore, this team has a pace factor of 79.5
This is a fine way to figure out the pace at which a team plays, but the big issue which Haralabos takes with this is that it doesn’t account for how slowly another team plays against you. On average, an opposing defense does not have a significant affect on an offenses’ time of possession in an NBA game.
This is where Time Per Possession comes into play. Time Per Possession (I’m just going to call this TPP going forward) is calculated by just taking the average time an offensive team has possession. This stat is agnostic of the opposing teams offense (looking at you Memphis).
I grabbed some Per Possession data from inpredictable.com and did some basic calculations on them in Python. Uploaded all my code to GitHub so feel free to take a look below. I plan on doing some scoring analysis in the future so feel free to check out pace.py if you are interested to see what I am looking into there.
To start with, I am pulling in each teams offensive TPP from inpredicatable. The calculation I am doing to get the final pace is relatively simple:
Pace= (Seconds In A Game / ((Home TPP + Average TPP)/2)) / 2
I paired up all of the teams with their pace calculated by TPP (tpp pace), original pace and the difference between the two. This results in the table below:
Conclusions
Most of the table shouldn’t be too shocking. The Warriors have a significantly higher TPP Pace than normal Pace (just like what Haralabos was saying in his tweet). The main thing I think this can be attributed to is teams playing at a much slower pace against them since that is the general consensus of the best strategy to beat them. Alternatively, the Pistons seems to play significantly slower than their opponents on average which is the cause of the -1.75 difference.
Overall, for most teams it seems that TPP Pace has a negligible difference to original Pace, but I do think this is an interesting thing to follow as the season progresses and the sample size increases.
