As the companion thread to the one about the consistency of each player on our team, here's one that compares Derrick Rose to other big minute point guards in the NBA. To make the comparison more interesting I also included Derrick Rose's MVP year and Derrick Rose from just 2016 (when he started to "get it") to see how each of those versions of Derrick stacked up. In order to filter players I made the comparison between point guards who average more than 28 minutes per game, have a usage over 20%, and have played in at least half their games this year. Players who made the cut were Steph Curry, Russell Westbrook, Kyle Lowry, Chris Paul, John Wall, Mike Conley, Damian Lilliard, Jeff Teague, Isaiah Thomas, Kemba Walker, Kyrie Irving, and Reggie Jackson. First was an average of their game scores. This was used as a proxy for PER since it's not available on a game-by-game basis. The whiskers represent the standard deviation in each player's output: Next up is the standard deviation of each player's Game Score: Looking at the standard deviation all by itself can be misleading because a Game Score is not rate-based and so guys who play more minutes will have it look worse than it really is. However, this graph probably isn't as problematic as the one presented in the last thread since all of these guys are big minute players. Next up is their median absolute deviation, which IMO does a better job of capturing someone's consistency because it'd remove abnormally hot nights as well as times when guys are playing through injuries which might decrease their stats. It measures how far guys typically stray from their median performance, not their average. Results: However, the MAD, like the St Dev, is based off someone's Game Score, which isn't rate-based. So ideally you'd want to adjust a players MAD according to their output to make sure it's proportional. When you do that these are the results you get: I think the above graph is probably the best representation of how reliably guys can "bring it" night in and night out. The bad news is that Derrick Rose is the worst guy in the group by a decent margin. The good news is that 2015 Derrick is near the top of the pack. One more stat we can use to measure consistency is skew. Skew is a measure of whether or not your distribution is top-heavy or bottom heavy. If you're bringing your average up with unusually good performances it'll be positive. If you're bringing it down with unusually bad performances it'll be negative. A perfectly symmetrical distribution will have a skew of 0. Together I think MAD% and skew are probably the two most reliable ways to gauge how consistent a player is. To me it's interesting that the best players are also the most consistent (more or less). I'd guess consistency is a unique proxy for ability. The more consistently you have an advantage over your opponent the easier it is to hit your average numbers. If you were to extrapolate from that, the take home lesson might be to not get sucked in by the promise of try-hard hustle guys who have little natural ability. I'd guess CamBear is going to be a wildly inconsistent player throughout his NBA career. Not because he isn't reliable, but because it's too easy for him to get stonewalled by the opposition due to his modest ability.
Also, to build on the "consistency = talent" idea, in his MVP year Rose's median deviation was more elite than his raw production, although both were very good.
