clock menu more-arrow no yes

Filed under:

The Mets are doing a good job of managing Harvey's workload

New, 55 comments

It's okay for the Mets to let Matt Harvey pitch past the sixth inning, as long as they keep him on a responsible pitch count.

Andy Marlin-USA TODAY Sports

No Harvey Day would be complete without its share of intrigue, buzz, or controversy, and Matt Harvey’s start against the Yankees had plenty of each.

The controversy du jour was how deep into the game the Mets let Harvey pitch. The right-hander had thrown 91 pitches through eight innings, allowing just two runs, four hits, and one walk, while striking out six. The Mets had a comfortable 8-2 lead, but given that Harvey was making just his fourth start removed from Tommy John surgery, some were surprised to see him take the mound for the ninth.

After retiring Brett Gardner and Alex Rodriguez to start the inning, Harvey allowed a single to Mark Teixeira and then walked Brian McCann on four pitches. It was then that manager Terry Collins pulled his ace and brought in Carlos Torres to record the final out.

Collins caught a lot of grief for keeping Harvey in the game as long as he did. Newsday reporter David Lennon, for example, scolded Collins for his lack of "discipline," and warned that he will "have to resort more to tough love in the future."

This criticism, however, seems misplaced. Instead of judging Harvey’s workload by how many innings he pitched, Collins used the much more reliable indicators of pitch count and stress of pitches, which were both quite low. Once Harvey showed signs of fatigue, Collins gave him the hook. This is a textbook example of how pitchers should be managed, and consistent with how the Mets have handled Harvey so far this year.

In a 1998 Baseball Prospectus article, Rany Jazayerli explained the importance of pitch count over innings pitched. To make his point, he compared the Dodgers’ handling of Orel Hershiser to the Braves' handling of Greg Maddux. In 1989, Hershiser led the league in both innings pitched (256.2) and batters faced (1,047); it was actually the third consecutive year in which Hershiser led the league in the former category. In 1993, Maddux similarly led the league in innings pitched (267.0) for the third year in a row, as well as in batters faced (1,064).

Judging by those numbers alone, the two pitchers appear to have had very similar workloads. Here’s the difference: in 1989, Hershiser threw at least 130 pitches in five different starts. In his final five appearances of the year, the Bulldog made starts in which he threw 125, 134, and an absurd 169 pitches—on a fourth-place team that finished 77-83, mind you. Hershiser was never the same after that season, pitching only four games the following year and never again approaching his dominant late-eighties form.

Maddux, by comparison, hit the 130-pitch plateau exactly zero times in 1993. Jazayerli notes that Maddux threw "fewer pitches per batter, and per inning, than anyone of his generation." So, even though the Braves’ ace threw a ton of innings in 1993 and in previous years (averaging over 260.0 innings per year from 1991 to 1993), he was hardly overworked. As a result, Maddux continued to be an extremely effective pitcher for another decade.

Baseball Prospectus research finds that throwing too many pitches in a single start is far more dangerous than throwing "too many" innings—either in a start, or over the course of a season. The reason is that throwing a lot of pitches in one outing is the greatest cause of arm fatigue, and pitching with arm fatigue is the greatest cause of arm injuries. As Jazayerli famously put it, "Throwing is not dangerous to a pitcher’s arm. Throwing while tired is dangerous to a pitcher’s arm."

Interestingly, BP’s research suggests that pitchers can make more starts and throw more innings per year without putting themselves at greater risk of injury, as long as their pitch counts are limited in each appearance: "There’s no medical evidence that four days of rest are better than three for allowing the pitcher’s body to recover," BP finds. Nor is there any "evidence to suggest that starting on three days’ rest hurt[s]…pitchers’ performance." BP concludes that, "by pulling starters more aggressively as their pitch counts and fatigue levels rise," teams can actually use their best pitchers more frequently during the year, and perhaps even "see a return of the 275-inning season."

In short, rather than use an innings limit, it would make far more sense to put Harvey and other recovering pitchers on a pitches-per-start limit. Allowing Harvey to make 32 starts and throw 200 innings is not inherently dangerous, as long as he isn’t overworked in any single appearance.

This is what made the Nationals’ decision to shut down Stephen Strasburg for the 2012 postseason so puzzling. As Jazayerli explains, the Nationals had skillfully managed Strasburg’s pitch count in each of his starts that year. Therefore, "the risk they [were] trying to mitigate had already been mitigated." There is no medical evidence suggesting that the Nationals made Strasburg any safer or healthier by abiding to some predetermined artificial innings limit. Their decision to bench Strasburg in the playoffs was simply unnecessary and may have cost them a shot at a championship.

And yet, three years later, you’ll still find those calling for Harvey to be put on an innings limit. In his column, Lennon argues that the way Collins handled Harvey against the Yankees "can’t continue in these types of scenarios. Harvey is averaging 6 2/3 innings through his first four starts, and if you extend that to 30 starts, that's a projected total of 197 innings for the regular season alone—or seven more than the upper ceiling of his limit." Again, no mention of Harvey’s pitch count of just 91 pitches going into the ninth; only that another inning of work would get him closer to that magic number of 190 innings by year’s end.

While pitch count is an imperfect way to measure arm fatigue, Jazayerli invented a system of Pitcher Abuse Points that roughly quantifies the stress level of each pitch thrown. In each iteration of his system, Jazayerli finds that high stress levels tend to kick in at around 120 pitches. Starting at 120 pitches, managers should tighten the leash and keep an especially sharp eye out for signs of fatigue. Pitches zero to 100 constitute minimal-stress pitches, while pitches 101-119 cause low-to-moderate stress.

Another possible way to measure stress level is by a situation’s leverage index. Although there hasn't been much research on this, it seems intuitive that pitching with runners on base in a close game would be more stressful and labor-intensive than cruising through a blowout game with the bases empty would be. Between pitch count and pitch leverage, we seem to have two reasonably good ways to measure a pitcher’s stress level in a given start.

With all of this in mind, let’s revisit Harvey’s start at Yankee Stadium. After eight innings, Harvey’s 91 pitches translated into a PAP score of zero, indicating a minimal level of arm abuse. Moreover, he had pitched in extremely low-leverage situations throughout the day. Before the ninth, Harvey had retired 16 of the previous 17 batters he faced, and the Mets had held no less than a four-run lead since the fourth inning.

In the bottom of the eighth, Harvey retired the side in order on 10 pitches, including two over 95 mph against the final batter of the inning, Garrett Jones. There is nothing to suggest that Harvey was experiencing a stress level that warranted taking him out of the game.

Once he did show signs of fatigue—namely, by giving up a hit to Teixeira and walking McCann on four pitches with two outs in the ninth—Collins removed him. Harvey threw a total of 107 pitches. Even after a bumpy ninth, Harvey finished with a still-low PAP score and, by a comfortable margin, turned in his least stressful start of the season by leverage index.

Keep in mind that this was Harvey’s only start of the year in which he threw at least 100 pitches. His pitch counts in his other four starts were 91, 95, 84, and 93. Furthermore, among qualified starters, Harvey ranks 94th out of 109 in average leverage index, and 104th out of 108 in high-leverage pitches thrown.

Part of this, of course, is due to the fact that Harvey is a great pitcher and doesn’t allow many runners to reach base. But part of it to be sure is Collins being conservative with Harvey’s pitch count and keeping him out of potentially high-leverage situations late in games.

The one red flag to keep an eye on is Harvey’s performance after 100 pitches. So far in his young career, it hasn’t been good:


Unfortunately, Baseball Reference only breaks these numbers down in 25-pitch increments, so we don’t know exactly where the drop-off in performance is occurring. But what’s clear is that, somewhere after 100 pitches, the wheels fall off for Harvey. For the most part, that's probably due to hitters getting more comfortable after seeing him a few times; as you can see, hitters tend to crush Harvey the fourth time through the order:

Time through the order PA BA OBP SLG
First 368 .203 .240 .304
Second 368 .173 .238 .267
Third 300 .232 .283 .315
Fourth 31 .464 .516 .500

However, it’s possible that Harvey’s struggles after 100 pitches are also the result of fatigue. Given that possibility, and how dangerous it is to pitch when fatigued, the Mets should be very cautious with Harvey after he reaches that triple-digit pitch plateau.

Going forward, the Mets would surely be happy to get only 100 dominant pitches per start from Harvey. But, long-term, it may be beneficial to both Harvey and the Mets to gradually and carefully stretch him out, so he can learn to pitch effectively after reaching that 100-pitch mark. If that’s the goal, his start against the Yankees is a good blueprint for how to do it: namely, in low-leverage situations, and with a short hook after 100 pitches.

By keeping Harvey’s pitch count low—and only modestly expanding it in low-stress outings—the Mets are managing their ace responsibly. If they stick to this formula, Harvey should be healthy enough to pitch and thrive in October, as well as in the years to come.