The last post was meant to show that method of projection has some sort of validity over the long run – but it’s useful for things other than hindsight and moaning about what could have happened in a season. Although it gets more prone to error over the short term, we can also use it to estimate a pitcher’s record based on their ERA and run support. This has a few uses:
1) Determining whether or not a pitcher “just knows how to win”.
You know the old chestnut- a certain pitcher might not pitch that well all the time, but he’s able to do just enough to get the win. Jack Morris is the archetype; the Blue Jays Version is Gustavo Chacin, with his career record of 25-15 despite a 4.18 ERA.
Unfortunately for the little person inside you who has watched too many movies and just wants to believe in gutsy performances and brave stands, this phenomenon never holds up to analysis. Gus allowed 93 runs in his career year of 2005, and was given a stunning 140 runs in support. Using these numbers, he was projected to win .693 of his decisions that season, which would have given him a record of 15-7. He actually finished 13-9. In other words, he actually figured out how to win less than he would have if the runs he allowed were randomly distributed that season.
2) Isolating run support vs. dumb luck
We all know that a pitcher’s wins are a pretty useless way of determining how they’re performing, but they’re never going to be abandoned because they provide a broad, easily digestible at the season so far. But we can adjust them to isolate two factors:
- Dumb Luck: by comparing the pitcher’s actual wins to how many they were projected to have (based on the same runs scored and allowed), you get an idea of how much they have been helped or hurt by the distribution of said runs (which is for the most part random). Lets call that EWins for “expected”.
- Run Support: Similarly, comparing how many wins a pitcher would get with the team’s average run support to the number of Ewins he got with the run support they gave him (I’m using Ewins instead of actual wins to eliminate the luck factor), you get an idea of what sort of effect the support or lack thereof of the team has had on a pitcher. Let’s call that N wins for “normal”.
Here are the Blue Jays Starters this year, adjusted for luck and run support:
I really like Nwins. Saying that Roy Halladay has received an extra run and a half per game doesn’t really mean much to most people. Saying that because of that he’s won about 3 games he would have otherwise lost does. Of course this is particularly useless for the guys who haven’t had a lot of decisions (especially Marcum), but it does show the extent to which Roy Halladay has been bailed out this season by the offence, and the fact that despite the sensation that AJ has been pitching better than his record, his W-L is right where it belongs.
3) Figuring out how well a pitcher would do on a better team
Let’s take Dan’s example of Matt Cain from a previous post (check out the blow-by blow of his torturous season in the comments). He is now a ridiculous 3-10, but his losing record is not so much the fault of the Giant’s woeful offence as it is with luck and him getting runs at all the wrong times. His record should be .500 this season because he’s allowed as many runs as he’s received in support. It would rise to 7-6 if the Giants scored their average number of runs for him, but it would only make it up to 8-5 if he played for a league average team (like the Blue Jays).
Going 3-10 has only been possible because of the insanity-inducing pattern the runs have been scored in, and that won’t last in the long run. But still – Matt Cain right now is a .500 pitcher for the Giants, but would be in line to win 18 games for the Tigers.