People have been using Kyrie Irving's true shooting percentage (TS%) as support for the fact that he's the undisputed top prospect in this year's draft. There was a discussion yesterday between Oceanary and Eric in Madison regarding the extent to which small sample size affects Irving's TS% (he played 11 games, 8 in the regular season and 3 in the tournament).
It occurred to me that Kemba Walker was off to an awesome start this year as well. So I use him as anecdotal evidence of how TS% can change over the course of a season.
This first figure compares TS% between Irving and Walker over the first eight games (before Irving got injured). The statistics are cumulative. Walker gets off to a slower start, but catches up to the point where he's actually better than Irving after six games, then falls slightly below Irving by the end of the eighth game.
This next figure shows what happens next. While Irving is hurt, Walker hits a slump. His TS% drops steadily until it settles in at the 0.550 range by about the 20th game. According to statsheet.com, Walker's season-ending value does not place him in the top 100 NCAA players who played at least 26 games.
This is what reversion to the mean looks like, folks. So what's Irving's "true" mean? That's the big question. He came back for the tournament and put up three more games that looked just like the eight regular season games, so maybe he's just that good. But eleven games is a very small sample. Looking again at Walker's season, his 11-game moving average TS% values range from 0.443 to 0.663. That's all the same player, in the same season, putting up that range of results over different sets of eleven consecutive games.
I'd still take Irving first, but teams will have to look at him a bit closer to decide whether he's as good as he appeared to be during a relatively brief stretch of games.