I'd like to propose the creation of a new statistic, an accounting stat for relievers that might help separate out a little of the noise from the generally awful save statistic. The new figure won't have much, if any, predictive value, but would merely exist for what I call "accounting" purposes -- a record of what happened for the curious or trivially-inclined, but of little use for looking ahead. It's for fun, and I deny any responsibility for harm that comes to you or your loved ones for unapproved or unauthorized usage.
I hate the save.
I do. It's a number whose existence implies meaning that isn't there, a reckoning that has altered bullpen usage, roster construction, and the game's salary structure, thanks to two generations of managers, pitching coaches, and front-office executives who grew up hearing relievers praised for high save totals and others excoriated for their apparent inability to handle the pressure of the ninth inning.
The save falls into the worst category of counting statistics, those that depend on context as much as -- or more than -- they depend on the individual player's performance. The pitcher win is a good example, while the RBI is the Laurent Gbagbo of baseball stats, refusing to bow out gracefully even when its time has clearly come. (I promise to give the RBI a fair trial and painless execution when its day of reckoning finally arrives.)
The save gives a performance by a pitcher additional weight based on the score and inning, rather than based on the pitcher's actual performance. Consider: Three outs in the ninth with a three-run lead gets you a save, but the same three outs recorded in a tie game, which is higher-leverage work, do not.
That example gets at a minor criticism of the save -- that is, a pitcher can pitch poorly and still get one. With a three-run lead, a closer can give up two runs in his inning of work and still notch the save, even though I don't think many analysts would consider the 2:1 runs allowed-to-innings ratio an acceptable one. Yet we have no line in our ledger to reflect these pseudo-saves, to show that while the reliever fiddled, Rome reached medium well before the fire was finally extinguished.
With the glorious Play Index at baseball-reference.com, we can, in fact, see who the primary culprits were in collecting these ill-gotten saves (as you can see in the top chart at right), and I'd like to propose that we name the non-scoreless save after the man who earned more of them over his career than anyone else did. I give you my new statistic, the Hoffman.
"But hey," you say (or would say, if I was secretly controlling your voice), "you can give up a run in a two-inning save and it's not that bad, really -- okay it's not that great but it's just fine, I mean, we can live with that can't we?" Sure we can! So let's change the criteria to one run or more allowed in no more than one inning of work, again with a save "earned". See the second chart.
That's right: Trevor Hoffman earned 27 percent more of these "short Hoffmans" than any other reliever in the history of baseball. Of course, that's partly a function of Hoffman pitching as long as he did, and earning more total saves than anyone else (a situation likely to change within the year), so let's look at the top ten leaders in saves with their Hoffman percentage (Hoffmans/total saves):
The most interesting part of that table, to me, is that there's someone below Mariano Rivera in percentage -- Wagner beats him by a Hoffman, since if Wagner had earned one more he would have been fractionally higher in the percentage column.
The range of Hoffman percentages among save leaders isn't all that high -- if you earn a lot of saves, you'll earn a lot of Hoffmans along the way. You can separate yourself slightly from the pack in this stat -- Rivera, Hoffman, and Wagner are the three best modern/one-inning closers, pretty much any way you slice it -- but the difference between these guys and average (10.7 percent) of the others in the top 10 is about a Hoffman every other year. That's not much.
The save is a bad stat, a point the Hoffman only further emphasizes.