May 14, 2010
Wrighting the Wrong
Apparently the talk of the town in New York these days is David Wright’s incredible strikeout rate. A sampling of recent headlines about the Mets:
David Wright is striking out at a prodigious rate, to be sure—he’s K’d in nearly a third of his plate appearances so far this season. (Incredibly, he still isn’t leading the NL in strikeouts—that dubious honor belongs to Justin Upton of the Diamondbacks. And another Diamondback, Mark Reynolds, is in third place.)
Of course, this isn’t the first time Wright has baffled people with his stat line. Last season, of course, was the incredible power outage—only 10 home runs all season, after putting up 30 home-run seasons pretty consistently before then. (Wright seems to have recovered his home-run stroke, at least, belting out seven so far.) So one year a power outage, one year a ballooning K rate.
What’s funny is that looking only at results, not much seems to change for Wright:
Neither a lack of power last season nor a high K rate so far this season have kept Wright from being an outstanding hitter. Notwithstanding all the headlines, Wright has been the Mets’ best hitter. Looking at WARP, he’s already put up nearly as much value this season as an average position player does in 150 games.
But what about down the road? Is Wright’s increased whiff rate a harbinger of things to come?
Wright’s K rate so far has been .299 in 144 plate appearances, compared to a PECOTA projection of .187. How likely is that result, assuming nothing at all has changed for Wright—that is, assuming that it’s totally a fluke? (And while we’re at it, assuming that the PECOTA forecast was totally on the nose—more on that assumption in a minute.)
OK, so strikeouts divided by plate appearances is a binomial—either you struck out or didn’t. We can estimate random variance of binomials in a straightforward fashion, and taking the square root gives us the standard deviation. So, given his PECOTA projection and his PAs so far:
In other words, assuming a wholly accurate forecast of Wright’s “true” strikeout rate, in 144 plate appearances we should expect to see a strikeout rate within .037 of his forecast about 68 percent of the time.
Taking the difference between his projection and his results and dividing by the expected random SD tells us that Wright’s strikeout rate so far this season has been a little more than three standard deviations away from his projection. We should see outcomes like that roughly one-quarter percent of the time.
That said—there are, so far, 196 batters with at least 100 plate appearances this season. Hitting on a quarter of a percent change in 200 tries is really nothing to get worked up about. (Especially when you consider we could just as easily have looked at home run rates, walk rates, hit rates, etc.—add up all the permutations and the odds of something being three SDs or more out of sorts go from “outlandish” to “nearly certain.”)
And we picked on Wright because we knew something weird was going on. Call it “selective sampling” or “cherry picking,” if you will. From a statistical point of view, we can’t call what we’re seeing here significant, in spite of the magnitude of the effect, because we plucked him out of a larger population.
That doesn’t mean we can rule out the idea that something has changed with Wright. But it means that we don’t really have any evidence that something has changed. Our most likely supposition is that this is just one of the many, many flukes that occurs over the course of a season of baseball.
And of course, even if he does keep striking out at this rate, so long as he keeps doing everything else at the same clip then he’s still one of the most productive hitters in the league. So don’t worry, Mets fans—at least, not about Wright. (It’s probably still OK to worry about the rest of the roster.)