Baseball Prospectus has a handy dandy little statistic called WARP, which behaves like WAR does everywhere else. But it’s their version of the stat, and you know it’s theirs because of the P. Anyway, it is a great tool because, just like WAR, it shows you the overall value of a player over a replacement player. Unless, of course, that player is John Lannan.
Yesterday, Jay Jaffe of Baseball Prospectus tweeted this:
John Lannan has been worth 0.8 WARP combined over last 2 yr (0.5 over last 3). At $5 mil/win, he’s a $2 mil player with $5 mil salary. Ouch
Very ouch, indeed. But that would make it seem that Lannan wasn’t really a valuable player. And I find this hard to believe. Not because I watch him and like what he does, or because he seems nice, or he’s been a long time National, but because other numbers say he is indeed valuable.
On Monday, Jaffe wrote an article for BP where he talked about the top rotations in the league, and part of that analysis included Lannan’s stinky WARP. I wrote a little about it here but I wanted to explore the strange results for Lannan a bit more. In the comments section of the article, I wrote that his ERA+ doesn’t mesh at all with his WARP. Jaffe was kind enough to respond to me:
The answer to this and most of the WARP/ERA disjunctions is that WARP is driven by Fair Run Average, which accounts for the pitcher-independent outcomes (he’s got a low K rate and a mediocre walk rate), adjusts for defensive and bullpen support, and credits sequencing (the ability to get a double play grounder or to avoid a bases-loaded walk, at the extremes). Lannan is above 5.00 over the past three years while barely being above replacement level according to that. I think that’s underrating him a bit, but WARP and PECOTA aren’t so easily impressed.
Ok, I get it, ERA+ shows what happened, and WARP showed what should have happened. And Jay (who, by the way, I think is a great writer and isn’t the boss of PECOTA, he just uses the tool and has a more intimate knowledge of it than me) not only suggested that it does indeed underrate Lannan, he was again nice enough to respond to me via a tweet, saying “PECOTA hates his BB and K rates, 3.3 and 4.5 over last 3 years.” But it still doesn’t work for me, because a) the implication is that John Lannan has been really lucky for 3 out of his last 4 seasons and b) the years with the worst WARP have better K/BB rates. And if you look at other WAR ratings, they aren’t so sure he’s super duper lucky. Look at what he’s done year by year (the red numbers in parenthesis are negative):
So looking at that, you’d think 2008, with its better K/BB rate and GB/FB rates would be the best. I can understand 2009 might have been his most valuable season, since he basically pitched 3-4 more games worth of innings. But I simply do not understand how 2010, with around 20% fewer innings, significantly lower ERA with respect to the rest of the league, lower K/BB rates (based on more Ks, and more walks) and a much lower GB/FB rate than 2008 was so much better. Looking at the numbers, it seems like 2010 was his worst season, and Baseball Reference and Fangraphs agrees. B-R sees it significantly worse, while Fangraphs has it very close.
It’s strange to me, and I don’t buy that Lannan has just been lucky every year, while in 2010 he was actually kind of unlucky. I think there is something that doesn’t get picked up by Fair Run Average, and that is that some guys are just better than the numbers would suggest. I don’t think this is an indictment of WARP, by the way. For the most part, it does a great job of looking at what players contributions were in a year. I think there are just a few outliers that don’t get properly assessed by this very robust statistical tool. It seems that John Lannan is one such player. For John Lannan, like Albert Einstein, WARP just doesn’t work.