Individual Scoring Variance
Recently I shared project that had been percolating for awhile, looking at team offensive and defensive not just by efficiency, but also by the variance in that efficiency. The idea is that most of the stats we look at are averages and while those can be great descriptors of what’s happening on the floor they do a poor job of demonstrating the highs and lows of performance. My last post was focused on the team level, but I also wanted to see what variance looked like at an individual level as well.
Hickory-High’s own Andrew Koo helped me sift through an enormous data set and collect a comprehensive digital pile of player game logs, from which we calculated individual scoring variance. The visualization below shows every player who scored at least 15.0 points per game since the 2004 season. Each player’s season is marked by their per game scoring average and their scoring variance, measured by the standard deviation in their per game scoring average.
There’s a lot to play around with here and a few really interesting things popped out to me:
- Kobe Bryant had the season with the highest scoring variance in our sample, 2007. This also happened to be the season of his 81-point game.
- Kobe appears in this group ten times, and all but two of his seasons had a scoring variance above the average mark.
- Carmelo Anthony also appears in this group ten times. All but three of his season had a scoring variance below the average mark. His three seasons with an above average variance were his last three seasons with the Knicks.
- Kevin Durant appears in this group seven times and in every one of those seasons his scoring variance was below the average mark.
- There were only two seasons in our sample where a player averaged more than 30.0 points per game with a scoring variance below the average mark – LeBron James in 2008 and Kevin Durant in 2010.
- LeBron’s 2013 campaign had the lowest scoring variance by any player in our sample who scored at least 25.0 points per game, and by a fairly significant margin.
- The lowest scoring variance by any player in our sample was Rashard Lewis‘ 2008 season in Orlando.
- Andrew also helped me out by running by running correlations between player variance and a few different factors in their game logs. I assumed that the two biggest stylistic indicators of variance would be three-point attempts (increasing variance) and free throw rate (decreasing variance). In actuality neither relationship proved to be very large. The r^2 for scoring variance and 3PTA was 0.103, for FTR it was 0.016.