Flu-Like Symptoms: Running Relievers Ragged

Last week, I introduced—OK, made up—the LIP Index as a way to measure wear and tear on relief pitchers. The LIP Index is simply the product of innings pitched and the average Leverage Index during a reliever’s appearances. It measures both the volume of a reliever’s work (innings) and the pressure he faces during his games (leverage). The LIP Index leaders last year were Sam Dyson, David Robertson, Steve Cishek, and Dellin Betances.

I also checked whether accumulating a high LIP Index portends future problems. Are hard-worked relievers more likely to break down? To find out, I looked at the 15 pitchers who compiled the highest LIP Index through the All-Star break last year. To my surprise, they posted a lower average ERA and a higher strikeout rate after the break. Only three of the 15 were unquestionably worse after the break. So in the case of relievers, at least, it appears that high-leverage workloads in the first half of the season don’t foretell problems later on.

Or maybe they do. I looked at pitchers within a season. As my editor, Aaron Gleeman, asked me, what happens year-over-year? Does a pitcher who sustains a heavy LIP Index workload in a season have problems the following year?

To find out, I compiled a list of the top 20 LIP Index pitchers for every season from 2003 to 2015. That gave me a group of 260 pitchers. For each, I calculated their ERA, SO/9, and BB/9. Then I checked how they did in the following season. Did their performance suffer as a result of their heavy, high-leverage workload the prior year? Did it make no difference? Or, as in the case of high-LIP Index relievers before and after the All-Star break last year, did they do even better?

When comparing the results from one year to another, 31 of the 260 relievers who had a high-LIP Index season didn’t pitch 20 relief innings the following year. Of those 31, I gave six a pass: John Smoltz in 2005, Miguel Batista in 2006, C.J. Wilson in 2010, and Josh Collmenter in 2014 all joined the starting rotation; Ugueth Urbina was arrested (and later convicted) for attempted murder after the 2005 season; and Mariano Rivera retired after the 2013 season. So there were 25 pitchers who truly failed to answer the bell the following year, due to injury and/or ineffectiveness.

And what happened to the remaining 229 pitchers? Here’s the rundown.

• They had an average ERA of 2.73 during their high-LIP season. The following year, their average ERA rose by 0.60, to 3.33.
• They had an average SO/9 of 9.42 during their high-LIP season. The following year, their average SO/9 fell by 0.27, to 9.15.
• They had an average BB/9 of 3.17 during their high-LIP season. The following year, their average BB/9 rose by 0.17, to 3.34.

Worse ERA. Fewer strikeouts. More walks. And this wasn’t the product of just a few pitchers posting terrible seasons. Excluding the six pitchers noted above, there were 254 relievers in my study. Of the 254, 154 had a higher ERA the following year, 133 had a lower SO/9, and 120 had a higher BB/9. Combining those with the 25 pitchers who couldn’t appear in even 20 innings, and 70 percent of the high-LIP pitchers regressed in ERA, 62 percent regressed in SO/9, and 57 percent regressed in BB/9.

So, pretty clearly, a high reliever workload in one season, measured by LIP, puts a pitcher at risk for worse performance the next year.

Except it doesn’t.

Here’s why. Over the winter, I wrote a number of articles in what I called the New Year’s Resolution series. I looked at players who’d significantly changed parts of their game—hitters who pulled the ball more, or exhibited greater plate discipline; pitchers who induced more grounders or improved their control; that sort of thing.

One of the ways I feel I’ve improved as a baseball analyst is through the comments that people like you leave for me. BP readers and subscribers are a smart group! I’ve gotten good suggestions and research ideas from you. And one of the best suggestions I got, in a comment in the first New Year’s Resolution post, is from noted sabermetrician Mitchel Lichtman, aka MGL: “Use projections as your baseline.”

It’s a simple suggestion, but it has particular salience here. Take, for example, Eric Gagne in 2003. He led the majors with a 156.4 LIP Index: 82 1/3 innings, 1.90 average Leverage Index. He posted a 1.20 ERA, 15.0 SO/9, and 2.2 BB/9. That’s pretty otherworldly. The following year, he had a 2.19 ERA, 12.5 SO/9, and 2.4 BB/9. A slip from his 2003, when he won the Cy Young award? Absolutely. But was his 2004 a disappointment? His ERA, SO/9, and BB/9 were all outstanding. For 2004, our projection system, PECOTA, expected a 2.07 ERA, 12.4 SO/9, and 2.5 BB/9. So Gagne was actually pretty much in line with our projections—a touch worse in ERA, a touch better at strikeouts and walks.

The advantage of comparing results to projection systems is that projection systems take outliers into account. If a player has an established level of performance, then markedly exceeds or falls short of it, the projection system will assume some regression to the player’s norm. A lot of the high-LIP Index relievers had amazingly good seasons, like Gagne’s 2003. Going forward, it’s not unreasonable to expect them to remain good. It’s probably unreasonable to expect them to remain amazingly good, though.

So with that in mind, I ran the same comparison I did before, but instead of comparing each reliever’s performance during his high-LIP Index year to his performance the following year, I compared his PECOTA projection for the following year to his actual performance. It told another story. Among the 229 pitchers:

• Their average projected ERA following their high-LIP Index season was 3.41. Their actual average ERA was 3.33, 0.08 better.
• Their average projected SO/9 following their high-LIP Index season was 8.86. Their actual average SO/9 was 9.15, 0.29 better.
• Their average projected BB/9 following their high-LIP Index season was 3.38. Their actual average BB/9 was 3.33, 0.05 better.

So while the 229 pitchers, on average, fell short of the numbers posted in their stellar high-LIP Index season, they slightly beat their projections. And even including the 25 pitchers who weren’t able to complete 20 relief innings the following year as failing to meet expectations, 48 percent of the pitchers beat their ERA projections, 52 percent beat their SO/9 projections, and 50 percent beat their BB/9 projections. Those are all coin flip-level probabilities, suggesting that relative to projections, nothing seems unusual.

In other words, if we look at projections rather than the outsized results a reliever may post in a high-LIP Index season, there is nothing here to suggest that pitching a large number of high-leverage innings in one season adversely affects a relief pitcher’s performance the next season.

Maybe it’s a matter of conditioning. Maybe it’s that innings totals are pretty tame. (None of the 260 high-LIP Index pitchers threw as many as 100 innings.) Maybe some pitchers really do thrive on the pressure. Or maybe it’s something else. What this article and last week’s do seem to show, though, is that there isn’t a lot of evidence of reliever overuse—at least as defined as a high number of high-leverage innings—in the contemporary game.