At New English D we’re among the those who wish to see the pitcher win removed from our baseball consciousness. It doesn’t measure an individual pitcher’s skill, but that’s how people use it. A pitcher’s won-loss record is about his performance, but it’s also about his defense, his run support, the other starting pitcher, and the other team’s offense. Also, luck, but I’m fine with luck.
Our most recent podcast covers the topic at length, but evidence and examples can do more to convince you about the flaws of wins than my rambling ever could. The catalyst for this post comes from something I discovered last night when contributing to Brian Kenny’s noble effort to #KillTheWin:
Matt Harvey (Go Heels!) is better in games he doesn’t win than almost every other pitcher in the league is overall. It’s time we get his back.
The rules are simple, these are The Nine best season by Wins Above Replacement (WAR) for qualifying starting pitchers who won fewer than nine games. In MLB history, there are 8286 qualifying seasons from 1901-2012 with 1187 finishing with fewer than 9 wins. These are the best.
9. Cliff Lee, 2012 Phillies
6-9, 211 IP, 3.16 ERA, 3.13 FIP, 4.9 WAR
8. Ken Johnson, 1962 Colt .45s
7-16, 197 IP, 3.84 ERA, 2.80 FIP, 5.0 WAR
7. Dutch Leonard, 1949 Cubs
7-16, 180 IP, 4.15 ERA, 2.71 FIP, 5.0 WAR
6. Bill Gullickson, 1981 Expos
7-9, 157.1 IP, 2.80 ERA, 2.11 FIP, 5.0 WAR
5. Al Benton, 1942 Tigers
7-13, 226.2 IP, 2.90 ERA, 3.07 FIP, 5.0 WAR
4. Steve Rogers, 1976 Expos
7-17, 230 IP, 3.25 ERA, 2.85 FIP, 5.1 WAR
3. Bob Welch, 1986 Dodgers
7-13, 235.2 IP, 3.28 ERA, 2.78 FIP, 5.3 WAR
2. Curt Schilling, 2003 Diamondbacks
8-9, 168 IP, 2.95 ERA, 2.66 FIP, 5.7 WAR
1. Nolan Ryan, 1987 Astros
8-16, 211.2 IP, 2.76 ERA, 2.47 FIP, 6.6 WAR
Wins generally correlate with good performance, but there are many cases in which good performances don’t result in wins and bad performances do. Pitchers can improve their likelihood of victory by pitching well, but they can’t guarantee it. Wins aren’t a completely useless measure of pitcher performance, but when we have so many statistics that are dramatically better, why should be place any importance on wins?
Here’s some evidence writ large. If we use Wins to predict three other statistics, WAR, ERA, and FIP, it doesn’t look good for wins.
|Adjusted R Squared||0.38||0.24||0.13|
What these numbers tell us is that 38, 24, and 13% of the variation in these numbers can be explained by variation in wins. Let’s give Wins the benefit of the doubt and pick WAR for the graph. There is a clear trend, but there is a lot of variation in WAR that wins can’t explain. The sample size here is over 8,000. You can be both terrible and amazing and achieve the same number of wins.
It’s time to #KillTheWin.