I get ESPN The Magazine every….I don’t know, it just appears. It appears because it’s cheaper to subscribe to the magazine and get ESPN Insider for free with my subscription than it is to subscribe to ESPN Insider. I get to read John Hollinger on the NBA and Keith Law on baseball, and ESPN gets the illusion that four a few bucks less I’m being advertised to by the corporations who pay for space in their mag. In truth after I flip through the mag, confirm there’s nothing of interest, I take it to the gym and toss it on top of the pile of magazines in the cardio room, to be read by some guy on a recumbent bike under the illusion he’s working out. I get five bucks off on my Insider, ESPN gets to advertise colognes and shit to recumbent bike guy.
This week the flip-through grabbed me, in a column called “The Numbers” by Peter Keating. Who is Peter Keating? According to his Businessweek bio he’s “a senior writer at ESPN The Magazine. For the last five years, Keating has written “The Biz,” a column that tackles financial issues from the fan’s perspective. Before coming to ESPN, Keating was a senior writer at George, where he covered national politics, and at Money. His work has appeared in publications ranging from Mother Jones to New York to Fortune. He lives in Montclair, New Jersey.” His Twitter bio claims he’s taking stats and analysis to ‘the next level’ so I assume I’m in for some good stuff here, especially since ESPN staffs guys like Keith Law and Dan Szymborski–with them on staff, if this is their Numbers Guy Who Takes It To The Next Level, I’m expecting brilliance.
Keating’s headline and sub-headline: “And the Winner Is…In the case for AL MVP, the backers of Mike Trout and Miguel Cabrera both get it wrong. But an innovative stat called WPA gets it exactly right.”
WPA is Win Probability Added. In a nutshell, it’s horseshit for gauging a full season’s play by a position player or starting pitcher because it’s the RBI of saber-y stats. It’s opportunity and context based.
A team has a certain likelihood of winning a game at any point in the game. Anywhere from a very very small almost-but-not-quite-zero amount in, say, the ninth inning when they are down to their last out and trailing by anything over 2 runs, up to a very very big almost-but-not-quite-100-percent amount in, say, the ninth inning when their opponent is down to their last out and trailing by anything more than 2 runs.
There are various win expectation percentages then for every scenario. Per Baseball Prospectus if it’s the fifth inning and there is one out and you, the home team, trail by one run and there are runners on second and third, your team has a 33.3% chance of winning the game. If you hit a double and the two runners score, you are now in the fifth inning with one out, ahead by one run, with a runner on second and your team’s chances of winning have risen from 33.3% (or 0.3333) to 72.5% (or 0.7250). Your WPA for that at-bat is 0.7250 – 0.3333 = 0.3917.
That’s essentially the same as giving credit for 2 RBI, which we’re well past, statistically and analytically speaking. The hitter has to come up at that situation, have the two men on, and produce. The hitter cannot–cannot, cannot, cannot–control coming to bat in that context. For the batter to control that he’d have to be able to control literally every other play in the game leading up to that event and he can’t. He’d have to be able to control getting the two men to second and third. He can’t. If you’re willing to believe that the hitter can control the context in which he bats, and that hitting a double with two men on base is something he did that is different from hitting a double with no runners on, then you’re free to believe that, as Keating seems to, but you are laughably, horribly wrong. A player can’t control when he comes to the plate–his actions along with the actions of the other 17 people playing the game and the manager making roster moves and the umpires making calls impact all of that.
What the player can control is the double he hit, the same as he can control that in every other plate appearance he takes that year. So the easy solution is to do what stats like WAR and wOBA and TAv do: neutralize the context and reward the player for hitting that double. Or making an out. Or walking. Or whatever the hitter does that he can actually control.
Mike Trout is a leadoff hitter who took the first at-bat in 138 of the 139 games he played this year. If he took 138 PA on the first at-bat of the game, that’s 21.6% of Mike Trout’s plate appearances that 100% of the time are being taken with no runners in front of him and a fairly 50/50 WPA environment. To make much of any change to the WPA, Trout would have to hit a solo homer. And even that’s not too big: if Trout hits a solo homer to lead off a game at home that is tied 0-0 when he comes up, the WPA changes from 0.4412 to 0.4722, or, not much.
When Trout isn’t leading off ballgames he’s typically hitting behind the worst hitters in the Angels’ order–guys like Erik Aybar and Chris Iannetta who sport .320-.335 on base percentages. When Miguel Cabrera comes to the plate he’s hitting behind Austin Jackson (.377 OBP), Quintin Berry (.330 OBP), and Andy Dirks (.370 OBP.) Cabrera’s coming to the plate with runners on base in front of him far more often than Trout.
For an easy look at that without speculating based on the OBPs of the hitters in front of them: Cabrera took 333 PA with runners on this year. That’s 47.8% of his total PA. Trout came to the plate with runners on base 214 times, only 33.5% of his total PA.
What’s hilarious is that Keating comes to the conclusion that Trout was better than Cabrera despite all the WPA talk. What’s double-hilarious is that Trout blew Cabrera away in a stat that is overwhelmingly biased toward Cabrera based on batting order alone. What’s triple-hilarious is that Keating, who writes a fucking column with ‘numbers’ in the title, disparages WAR, which is context neutral and a remarkably good measure of what a player did with each and every opportunity he had in a season, in favor of his context-dependent darling WPA. What’s quadruple hilarious is that Keating even writes “The traditionalists who love RBIs, however, have a point when they complain that WAR doesn’t measure situational performance.” The whole mess reads like “Hey, I’m as statistically savvy as the next guy but would you mind for a minute if I cherry-pick the absolute worst saber-y stat with which to measure a season’s performance, and insist that because it placates the mouth-breather in me who thinks RBI are awesome, I just go with that one?”
Yes, Peter Keating, I do mind. God fucking dammit, I mind. And ending your piece “..let’s be smarter about the case for why” might fly with some guy who picked up the magazine for the NBA Preview, but it won’t in a smart room.