Which statistics to look at early in the season
With the DBL season just underway at about six games per team, what are the statistics that can already reliable tell us something? In his article “Calling the Race” [1], Kevin Pelton of basketballprospectus.com takes a look at the NBA season 2010-11 and examines from what point in the season statistics align with the eventual outcome at the end of the season. What can we learn from Kevin’s article?
Here are some take-aways:
- The error in straight point differential is within two points of the eventual outcome after about 25 games into the season.
- Player performance tends to stabilize more quickly: this is simply because the number of observations is so much larger. For instance a player playing 28 minutes per game has had a chance to get about 60 rebounds, which makes his rebound percentage quite reliable.
- Historically, analysts have used three playing time benchmarks as cutoffs, namely 250, 500 and 1,000 minutes. In the DBL the 250 minute cut-off will already be reached in the next couple of games by some starters. I’ve added a playing time worksheet to the Excel analysis sheet [3], where you can check this out yourself.
- Player statistics stabilize at different rates, based on the considered denominators. Assists, steals blocks and especially rebounds occur so often, that they stabilize very quickly. Same goes for player tendency statistics, e.g. usage%.
- Shooting appears to be more volatile, so don’t draw any “hot shooting” conclusion too soon.
Footnotes
| [1] | http://www.basketballprospectus.com/article.php?articleid=2442 |
| [2] | http://www.basketballprospectus.com |
| [3] | Excel Analysis workbook |