Over the past few season, I’ve been trying to figure out a good way to rate players individually. At first, I came up with something on my own, but then I realized that other good metrics were already out there. First, I came across Dean Oliver’s methods, which he detailed in his book, Basketball on Paper. I was fine with his Offensive Rating method of measuring performance, except that it seemed overly complicated and at times counterintuitive. Details and challenges to intuition are not bad things by any means, but they made the stats difficult to comprehend.
After I’d been using those formulas for about a year, John Hollinger started working for ESPN’s Insider section to track his Player Efficiency Rating for the NBA. Since his numbers were becoming more widespread, I started to use that instead of Oliver’s rating. Other methods like eWins and Plus-Minus are out there, with Basketball Reference and 82games being fantastic resources for countless other APBRmetrics.
Why am I bringing all this up now? I can’t decide which metric to use anymore. I’m growing more and more frustrated by PER, which depends too heavily on a player’s usage rating for my liking. I’ve made some adjustments to it and called it the Usage Rate weighted PER on my individual stat sheets, but I’m still not particularly happy with it.
I think I’m going to start tracking Oliver’s stats again, since their more theoretical basis seems to capture a player’s contributions a little bit better than PER, even if they are more difficult to understand. If you’re familiar with sabermetrics in baseball, Oliver’s stuff is more like Linear Weights, while PER has more of a Runs Created feel. I’ll still track PER, since there is a benefit to being able to look at it and immediately know where a player stands compared to his peers (because he’s either above or below 15). I’ll probably roll out this change sometime this week, and I’ll revive the search for other good stats to explain more of what’s going on in the games. Nothing beats watching the action in person, though.
Update: The latest stat file can be found here. There are two new individual stats at the far right: UPER (Usage-rate weighted PER) and SUPER (UPER adjusted for schedule strength). Personally, I’m loving the SUPER acronym. Those are my own changes to John Hollinger’s PER formula to adjust for the fact that some players don’t see the ball as much, and thus have fewer chances to affect their PER. I present all of my variations of PER (now four of them) as a comparison.