Jim Caple and the pitcher win
ESPN’s Jim Caple just posted an article about the win statistic. This seems to be a response to Brian Kenny’s all-out assault on the win. Kenny is attacking the old stat as a grossly misleading, if not useless, statistic in measuring how good a pitcher is. I don’t really care if Kenny’s “kill the win” campaign gains steam or not; frankly, I don’t really care about wins (outside of my fantasy leagues where they count). I do think it’s outdated and doesn’t tell us much of anything. Matt Harvey has nine measly wins while having (pre-injury, obviously) a rookie season for the ages. Harvey still ranks second in pitcher WAR on Fangraphs (as of September 18- Kershaw, at least, will eclipse him before season’s end). Of course, it’s not Harvey’s fault he plays for the Mets (I suppose he could have refused to sign, like Elway and Baltimore or Cushman and Denver) but poor Harvey was enjoying the 13th lowest run support of all NL starters.
Caple admits to some shortcomings, but comes to the defense of the win:
Perhaps stat-heads would appreciate the win more if it was something else, though, something much more complicated and mathematically challenging. Maybe they would like it more if it included complex calculations that account for run support, adjusted ERA, advanced fielding analytics, WAR, stadium factors and humidity and was called tWIN.
I do love the idea that the win is simple and other sabermetric stats are, by definition, not. Caple talks about the “uncomplicated” win in an article that also discusses how on September 13, Cleveland starter Danny Salazar struck out 9 in 3.2 innings, but couldn’t get the win, because he didn’t go the required five innings. He talked about Drew Smyly vulturing Max Scherzer’s 20th win, after Smyly blew the lead. The rules on how wins are awards are full of inane loophools and requirements. Caple doesn’t even mention my favorite part of the win rule, which I’ll quote right out of the official MLB rules:
Rule 10.17(b) Comment: It is the intent of Rule 10.17(b) that a relief pitcher pitch at least one complete inning or pitch when a crucial out is made, within the context of the game (including the score), in order to be credited as the winning pitcher. If the first relief pitcher pitches effectively, the official scorer should not presumptively credit that pitcher with the win, because the rule requires that the win be credited to the pitcher who was the most effective, and a subsequent relief pitcher may have been most effective. The official scorer, in determining which relief pitcher was the most effective, should consider the number of runs, earned runs and base runners given up by each relief pitcher and the context of the game at the time of each relief pitcher’s appearance. If two or more relief pitchers were similarly effective, the official scorer should give the presumption to the earlier pitcher as the winning pitcher. [emphasis added]
The win isn’t a simple statistic just because it doesn’t have a mathematical formula. Like the RBI, there’s so many things that the player collecting it doesn’t control. That alone is why its value is so limited.
Caple boils down his argument to this with this:
Could the win be better? Sure. But one of the reasons I like the win is its simplicity. Despite its clear limitations, the win is a long-established and fun statistic that quickly tells us something about a pitcher — how many bad pitchers win 18 games in a season or 200 games in a career? — though by no means everything. Nobody is saying the win is the ultimate arbiter of anything for a pitcher. It’s just one of many stats for your consideration.
As I’ve already said, the win is not simple, and Caple makes the point himself. And he’s right, it’s one of many stats for you consideration, like paying $3 for a tin of Pringles on your next US Air flight is a food option; it’s not necessarily the best option for you to take.
As for players who accumulate big single season win totals or lots of wins over a career- does this mean much? In the days of complete games, when pitchers would rarely be lifted, the win meant something though still less than it seems lots of folks want it to. These days, with increased specialization and pitch counts and so on, it means even less. Christy Mathewson, to pick a random old-timey pitcher with longevity, averaged 8.67 innings per start for his ENTIRE 552-start, 17-year-long career. Mike Mussina is the closest modern starter in terms of games started to Mathewson, as Moose started 536 games. For his career, Mussina averaged 6.67 innings per start. Mathewson would need one out from a reliever, but he still needed the run support. Moose needed two innings and an out plus run support. To turn this a different way, Curt Schilling and Mike Mussina were both terrific pitchers in a big offensive era. Both were regular starters from 1992 until 2007. Schilling started 569 games to Moose’s 536, but Moose threw almost exactly 300 more innings (Schilling averaged 5.7 innings per start.) Schilling ended up with 54 fewer wins. Why? Mussina played for the generally good 90s Orioles while Schilling was pitching for the generally bad 90s Phillies. Similar service time, huge difference in win totals. Incidentally, Mussina is at 83.0 in rWAR and 82.3 in fWAR, while Schilling is 79.9 and 83.5. What seems like the better comparison? Wins, where Schilling is only 80% the pitcher Mussina was or WAR where they were equally good?
As for the specific win-totals Caple mentions, to counter that point I do need some fancy sabermetric, context-controlled stats, specifically FIP- and ERA- (each stat takes into account park factors and the league average numbers and compares them to other pitchers of that specific era; 100 is average, and the lower the number, the better. Someone who has an FIP- of 80 is 20% better than their peers) Lew Burdette was a solid pitcher from 1950 to 1967, mostly for the Braves. He collected 203 wins. He had a career ERA- of 101 and a career FIP- 103, both slightly below average. Denny McClain in 1966 won 20 games, and had an ERA 13% worse than the AL average and a FIP 23% worse. It’s actually not hard to find other examples of pitchers who were average or worse and managed to still win 18 or more games (here’s the custom search) but I suppose I can be discounted because I used ERA- and FIP-. Are any of these pitchers bad? No, none of them were Jose Lima for the Royals bad (incidentally, Lima had a 21 win season for the Astros in 1999. He was pretty good that year. I’m pretty sure there were a dozen or two pitchers in 1999 that weren’t named Jose Lima you’d have preferred on your team.)
I understand a desire for a simple number that can be understood quickly and tells us who was a good pitcher and who wasn’t. What escapes my understanding is this luddite position that Caple and others like him take, where they run towards this an archaic stat. I suppose there’s a bit of “it’s always been there.” But it strikes me a bit like someone in 1920 saying “we can’t get rid of the guys making horseshoes.” Horseshoes might be interesting, and occasionally might be useful, but they’re not longer a critical part of our world.