Exit Sandman?

Is Mariano Rivera Slipping?


Mariano Rivera's late inning dominance has been as crucial to the Yankees' success over the last ten years as any single factor. But perhaps more impressive still is the consistency with which he has gone about it. In the four rate statistics believed to be more or less defense-independent—strikeout rate, walk rate, isolated slugging, and ground ball to fly ball ratio—and even in the notoriously defense-dependent or “luck” driven batting average on balls in play (BABIP), Rivera has been remarkably dependable. Most of the variation clusters closely around the mean: Rivera usually strikes out between 18 and 25 of every 100 batters he faces, walks between 4 and 7, and gives up 2 to 3 times as many ground balls as fly balls. Even in the metrics where there is a noticeable skew, it tends to be toward superior performances.

But in 2007, Rivera had what many regarded as an “off year”. Indeed, across both conventional and unconventional stats, he ranked as no more than an average closer. Of the thirty-six pitchers in baseball last year with at least 10 saves, Rivera fared thusly:

He faced more batters and gave up more hits (many of them line drives, more on this later) than almost any other closer in baseball. When the same thirty-six closers are given standardized “scores” that measure their value over or under the positional average, Rivera comes in at a measly .40 above the mean (by comparison Takashi Saito and J.J. Putz topped the list at 10.69 and 7.28 above the mean, respectively). Unsurprisingly, Rivera also saw his ERA balloon to 3.15, the highest since his rookie season (and the first time its been north of 2.00 since 2002).


The most popular explanations of Rivera's down performance focused on two factors: age and misuse. The age argument seems unlikely on the face of it. While it's true that at 37, Rivera was the fourth oldest closer in baseball last season—behind Todd Jones, Trevor Hoffman, and Bob Wickman—it's also true that six of the top ten closers last season were 30 or over (including Hoffman at 40, and Saito at 37). When power pitchers slip, the signs usually start to appear around the age of 32. But it was at that age that Rivera's most dominant five year stretch began (Between 2002 and 2006, opponents hit just .213 and slugged just .285 off Rivera, while he averaged over 40 saves per year with a 1.83 ERA and a WHIP of .98). While age eventually becomes a factor for any major league pitcher, starters tend to collapse more rapidly than relievers, and Rivera's overall stability and dominance through his mid-thirties provides little reason to think his decline phase will be anything more than gradual.

The misuse argument is a bit harder to assess. Was Rivera left in too long in some games? Perhaps. When pitching in the eighth or ninth inning, opponents hit just .250/.291/.337 off Rivera. In extra innings of work the numbers went to .227/.320/.500. But these increases are normal signs of fatigue for any pitcher, and the only 2007 figure substantially disparate from Rivera's career splits in those situations is in extra-inning slugging percentage (Rivera's career SLG-against in those situations is .340). And this is hardly unusual, as over a short period of time even one extra-inning home run can drastically inflate SLG (especially since closers tend to leave and/or games tend to end after extra-inning shots).

Was Rivera not used regularly enough? Was he over- and/or under-rested for significant stretches, and did this effect his performance? Again, there is no compelling pattern here. In 2007 Rivera pitched his poorest on zero days rest (.294/.333/.451 against, with an ERA of 3.38), which is normal enough, but his ERA ballooned to 5.27 on three days of rest, before going back down with each subsequent day. His career splits are for the most part tightly clustered, and where there is substantial difference it appears basically random: Rivera posts his best career ERAs on two, three and five days of rest, and his worst on one, four, and six-plus days of rest. None of this supports the case for improper usage.

Did Rivera under-perform in non-save situations, as the common bit of closer psychology predicts? Somewhat. Rivera's ERA for the 30 appearances in which he received 'no decisions' in 2007 was 3.30, higher than his career mark of 2.24. But Rivera's ERA was up from career averages across the board—from a ridiculous 0.68 in saves to a slightly less ridiculous 1.06, and from 16.17 to 24.30 in losses—so it is hard to pin the change to pitching to the score. It is also important to note that earned run average is a flawed measure of the underlying quality of a pitcher's performance, especially for relievers, for whom one or two bad outings can be unfairly distorting.

So if there is no smoking gun that points to age or misuse, what exactly caused Rivera's poor showing in 2007? The likeliest answer is luck. Pure, dumb luck. More on why next post.