Jim Albright / the japanese insider
New Pitcher Projection Method
Frankly, I never wanted to do pitcher projections, but I didn't feel I could avoid it if I did the hitter projections from NPB to MLB. I'd never seen pitcher projections done, and I'm still unfamiliar with the workings of the other efforts out there to do such projections of pitchers. I thought I could get away with simply using the inverse of the factors for hitters, and so I tried that. The truth is, it didn't work too well.
I didn't have much of a clue on how to fix the pitching projections for quite some time. The first breakthrough I had was the realization that projecting runs based upon the runs (allowed, scored, or batted in) in one place to the other league when the leagues have such differing effects on the pieces which make up those runs just doesn't work. What I'm trying to say is that average is much less effected by the difference of the leagues than homers are. As a result, an Ichiro will retain much more of his value in the majors from NPB than a power hitter like Hideki Matsui will. If you try to use one run factor for both Hideki Matsui and Ichiro, that truth is lost in the run and RBI projections. If you project the run and RBI figures from how many runs each man projects to produce under a runs created formula, Hideki Matsui's RBI and runs scored will be reduced more than Ichiro's will, as will occur in real life. The same is true for pitchers.
The means for calculating pitching runs came from Bill James' component ERA. I tried to use component ERA on the existing conversions of hits, homers, walks, and homers, and it was quite clear that my conversion factors worked poorly, especially the ones for walks and strikeouts. A major problem, beyond even the fact these particular statistics do not seem to project quite as well as many of the others, was that the factors were based on the hitters that played in both leagues. This subset of hitters had a lot of slow power hitters, and few singles hitters. This created major distortions in the conversion factors.
At that point, the only thing to do was to do a comparison of the performance of pitchers
who pitched in both the majors and NPB. I used matched numbers of innings pitched (with a
minimum of 0.1 IP) for the pitchers in both settings. What I got from 169 pitchers were
14907.2 matched innings, each pitcher pitching the same number of innings in each league.
The results were as follows:
* = Japan to majors
From the runs, I calculated a won-loss percentage using Bill James Pythagorean Theorem, and gave each starter one decision for each 9 innings pitched. As before, for pitchers from 1990 on, I allowed their innings pitched to rise in the projection, as innings pitched in Japan had dropped to levels similar to what one would expect in the majors in seasons as short as those played in Japan. Reliever won/lost totals were left alone, as the way the pitcher is used has such a dramatic effect on won-lost records for relievers that I thought it best not to mess with it. Saves were reduced if a pitcher's ERA rose over 3.00 by a very crude approach. Now I had all the means to try my new adjustment method, and I think it performs much better.
Now, it is time to show you how this method performs so you can decide for yourself. I'll do projections from the last five seasons of the main pitchers imported from Japan in the 1990's (so long as they played at least that many), then put a break, then show their first five seasons in the majors (again, so long as they played that many). After that, I'll make some comment on how I think the projection performed for that pitcher.
I'd say his projection is quite reasonable. My previous attempt at projecting his record can be found here
At least in 2004, he dramatically outpitched his projection. It will be interesting to see if he can keep it up. My previous attempt at projecting his record can be found here
It's a little hard to make the comparisons for him as he was used as a starter in Japan, and as a reliever in the majors. He pitched better than the projection, but I've got to wonder how much of the improvement was due to the change in role.
I'd say he pitched a bit better than his projection.
He wasn't quite as good as he was in the projected seasons, but he was quite valuable his first three years in the majors. My previous attempt at projecting his record can be found here
He was better than the projection in the first two years, though the fact he wasn't used as hard may explain it. After that, he was rather similar to the projections. My previous attempt at projecting his record can be found here
He didn't pitch as badly as his press from his stay in the majors would indicate. However, according to the projections, he seriously underachieved. I'd say the projection captures an essential fact about Irabu: he had that kind of talent, but his emotional makeup was such that going to a high pressure situation with the Yankees on top of the culture shock was more than he could handle. Therefore, he underachieved. My previous attempt at projecting his record can be found here
I'd say the projection did well in his case. My previous attempt at projecting his record can be found here