The sports stats world is a-twitter over a piece by Jonah Lehrer at the new Grantland site. Essentially, he says that stats have become important enough in sports that people are starting to ignore the immeasurables; it isn’t that the stats aren’t doing well enough, it’s that they’re doing too well (side note: you can find many replies to the article linked through stathead, and a reply by Jonah himself at wired). One piece of Jonah’s argument that hasn’t been addressed is his example of JJ Barea. According to Jonah, all the stats say JJ shouldn’t have been in those final three games of the Finals, but there he was leading the Mavs to victory. How true is that?
Jonah argues that JJ was put in the starting line-up for psychological reasons, not because any stats pointed to it being a good move. The first question you might ask is, where did his extra minutes come from? Even if JJ were a crappy player, playing him more minutes would be a good idea if he replaced someone worse. Comparing games 1-3 to 4-6 we can see that while JJ played 27 more minutes, Shawn Marion played 22 less; Jason Kidd 5 more; DeShawn Stevenson 3 more; Jason Terry 3 more; and Peja Stojakovic 21 less (he never got on the court in the last three games).
Barea nominally replaced Stevenson in the starting line-up, but DeShawn’s minutes didn’t change. Since JJ is a shooting guard, he could only really only replace another guard or a small forward (assuming the Mavs went to something like a three-guard line-up). So what we really have, roughly speaking, is 38 minutes of JJ, Kidd, Stevenson, and Terry replacing 43 minutes of Marion and Stojakovic. So let’s look at what the stats say about Marion and Stojakovic.
In 2011, Marion was either pretty good (if you believe Wins Produced or ezPM) or pretty bad (if you believe APM or RAPM). In the first three games of the Finals, he scored 46 points on 38 shots while getting 7 boards, 3 assists, 1 block, and almost 1 steal per game. In short, he played pretty well but the Mavs lost two of those games. Peja, on the other hand, was seen as average to good (RAPM was fairly bullish, but WP, ezPM, and APM also had him at a bit above average). But in the first three games of the Finals he was terrible, going 1 for 5 with no assists or blocks and only 3 total rebounds and 2 total steals in his 26 minutes. I don’t have the play-by-play data, but he was also presumably eaten alive by whoever he was guarding on defense (that’s what I learned by watching TV, where people who ‘watch games’ told me what was happening). In sum, it isn’t unreasonable to see either Marion or Peja losing minutes from a stats perspective. Marion was arguably a bad player in the regular season, and Peja was not contributing at all in the Finals through the first three games. Plus, the Mavs were down 1-2 and the coaches might have felt it was time for a change.
So what about JJ? As Jonah notes, he wasn’t exactly setting the world on fire with his 5 for 23 shooting in the first three Finals games. In the regular season, he was probably about average (average according to RAPM, above according to ezPM, a bit below according to WP, and bad according to APM). But Jonah also mentions that he was missing a lot of layups. That implies that he was getting to shoot a lot of layups, and layups are a pretty good shot. Unless JJ was missing because of a lot of shot-block pressure, there’s a good chance that he would make those shots as usual in the future. That is, of course, if you believe in the statistical idea of regression to the mean.
In short, the statistical viewpoint from WP, RAPM, APM, and ezPM for the regular season says that Barea was probably a better option than Marion and not a big drop-off from Peja. Perhaps most importantly, according to Mavs stats guy Roland Beech’s site, Barea had a better rating than Kidd, Marion, and Stevenson anyway. And in the first three games of the Finals, Peja was playing poorly in a low number of minutes while the Mavs lost twice. JJ wasn’t playing great, but it appeared that he had easy room for improvement. So I think you could make the statistical argument that JJ was in line for more minutes anyway, and I certainly think it’s defensible if you argue that the Mavs decided to make some kind of change after going down 1-2 in the series. Perhaps the Mavs staff was following the numbers after all.