In order to present my old study on relief pitching at SABR 36, I decided to update the stats and to also resolve an issue that cropped up in the original study. The problem is that I jury-rigged the stats for middle relievers by extrapolating between starters and old-tyme closers. The results were that, I felt, the evaluations for the best middle relievers were way out of skew.
Now that I have loaded Retrosheet's game log data, I can derive how well saving a run led to wins in different scenarios for different eras. I used the data for starting pitchers and closers to determine the scenarios that would apply to middle relief and to weight the various scenarios.
Here are the results for middle relievers by era. Note that they have not had much impact on the game until the late Seventies:
Runs Saved/Relief W
IP Per GS
I also listed the average appearance length for a middle reliever and for the starting pitcher for each era. Now, we can plug these numbers into my formulae for middle relievers, stir, and then we will have updated middle relief data.
I'll update all the stats to include the new middle relief numbers as well as all data through 2005 tomorrow.