Previous entries:

And The Winners Are…

OK, so now that we have gotten the preliminaries out of the way, we are ready to get to the results.

Here are the best reliever seasons of all-time based on relief wins. Also included are reliever types and Pitching Win Shares:

Now here are the worst:

Now here are the career best 100 relievers by relief wins, this time with the number of seasons per reliever type:

Which brings us to career worst relievers:

And just for fun, here are some other pitchers who either were mentioned throughout the study or are of some significance (including many of the best and worst active players and all of the Hall of Famers I could find. Note Bump Hadley is here because he is the most average reliever of all-time):

Validating the RWin Data Model

Well, now that we have results, are they meaningful? OK, Hoyt Wilhelm may be the best reliever of all time, but could Jim Kern's 1979 season be the best for a reliever ever? And even if it was the best ever, could have been that much better than any other?

There are some important things to remember. First, technically we are not looking for the best season but the most valuable one. And we define value by the pitcher who prevents the most runs above average weighted by the way the way that the pitcher was used. One could argue that a more recent reliever, say Gagne in 2003, had the best season ever, but given the way that he was used his potential value was lessened.

Second, Kern's 1979 was pretty darn good after all. He pitched 71 games, all in relief, had a 13-5 record, 29 saves, and a 1.57 ERA, 165% better than the adjusted league average. He threw 143 innings, allowed 99 hits, struck out 136, and walked 62. Total Baseball rated him the best player in baseball that year by Total Player Rating. He ended up fourth in the AL Cy Young vote in '79.

Besides, a number of stellar but forgotten seasons bubbled to the top of the list: Ellis Kinder's 1953 year, John Hiller in 1973, Mark Eichhorn in 1986, Dan Quisenberry's 1983, Radatz's first few years, etc. And the seasons that show up among the worst are truly awful and the ones that pitched the most innings show up as even more awful.

One other thing to keep in mind is that we are pigeonholing these seasons into a handful of reliever roles and assigning value accordingly. Kern fell into the Wilhelm category, the one for which the pitching runs saved are valued the most. If the criteria I used had been tweaked slightly he could have been placed in the Sutter group and his relief wins would have been slightly lower. But whoever would replace Kern at the top if we tweaked the criteria a bit would be someone already in the upper echelon. I think that the system is not perfect but that it points us in the right direction.

I feel that it hits the mark better than pitching Win Shares for relievers. I prefer that the average pitcher is at zero. In Win Shares a pitcher is sometimes given a fraction of a Win Shares just for showing up a pitching a few innings. So you end up with poor pitchers receiving either zero Win Shares or a half a Win Share or so. There's no way to differentiate a bad season from a truly awful one. With RWin the worse a pitcher is (i.e., has a high ERA in a lot of innings or gives up a ridiculous amount of earned runs in a smaller set of innings), the worse his final scored ends up being.

My one problem is with middle relievers. I feel that this system overvalues them, but I'm not sure how to rectify it. My brokered solution of splitting the difference between a Clint Brown type pitcher and a starting pitcher in order to calculate middle reliever relief wins does tend to lower the ratio of prevented runs-to-relief wins down in Hoyt Wilhelm territory when the innings pitched per game gets very low (well under one). For the most part the pitchers with these totals end up pitching so few innings that the relief wins are negligible. However, there are a handful (e.g., Quantrill in 2003, 4.62 Rwins with a 4.50 PR-to-RWin ratio) that get a bit out of hand.

To be completed…