Hot Stove U: Stress Pitches vs. Pitch Count

The Setup

On June 2 of last season, heading into the top of the ninth inning with the Toronto Blue Jays up 6-4 over the visiting Los Angeles Angels, Toronto manager Cito Gaston sent Roy Halladay back to the mound. Halladay already had thrown 116 pitches in the game.

Modern pitch-count orthodoxy would have had Halladay out of this midseason game at least 10 pitches earlier. So the question stands: Why would Gaston send him back out?

Obviously, Halladay is not some young pitcher who needs to be babied, but even so, 116 pitches is a lot. Why tempt fate with one of the game’s best pitchers and potential trade bait (with the trade deadline less than two months away) for a team that almost certainly would not be making the postseason?

In the end, Halladay closed out the game with 133 pitches, giving his team the victory. Did Gaston put Halladay’s arm at risk, or did he realize that not all pitchers are the same?

Those questions are relevant, but we’re here to demonstrate something else: Not all pitches are created equal.

The Proof

There is a growing belief that high-stress at-bats are more taxing than those in relatively low-pressure situations — and therefore that pitch counts from the two scenarios should not be treated the same. If a pitcher can breeze through easy innings in one gear and then kick it up to another gear when needed, raw pitch counts might not be the best tool to assess workload, either in an individual game or over the course of a season.

To test this theory, we need some measure of what we mean by pressure. We use a metric called Leverage Index, developed by statistician Tom Tango. Leverage Index (LI for short) quantifies the impact of every situation based on how the outcome will affect a team’s odds of winning a particular game. It is scaled so the average situation is always 1.00.

For example: An at-bat in the bottom of the ninth with two runners on and one run separating the teams will have a huge LI; the outcome of the at-bat will greatly affect the likelihood of either team winning the game. On the other hand, an at-bat in the middle of a 10-0 game with two outs and no runners on has a minuscule LI.

Let’s return to Halladay’s game against the Angels. Heading into the seventh, the Jays were up 6-0. Through those first six innings, because of the big lead and a dearth of Angels baserunners, Halladay faced just two at-bats with an LI of more than 1 and many with LIs less than 0.5. The 75 pitches Halladay threw through those six innings overwhelmingly occurred during low-leverage at-bats.

That all changed in the seventh inning, when the Angels managed four runs off him; as a result, his pitching changed drastically. He started throwing his curveball much more (19 times in 58 pitches in the seventh through ninth innings, compared to just 14 times in 75 pitches through the first six frames). It worked, as the contact rate on his pitches dropped from 72 percent to 60 percent and he struck out five batters in the final two innings, slamming the door and preserving the victory.

Those last 58 pitches likely were more taxing on Halladay than the first 75. With the game close in the late innings, Halladay shifted from pounding the zone with sinkers and cutters to get weak contact to throwing his breaking ball and trying to hit the edges of the zone to get strikeouts.

Halladay is not alone in shifting his strategy in high-leverage situations, although most pitchers respond by increasing the speed on their fastballs. In 2009, the average starter threw his fastball half a mile per hour faster in high-leverage situations. This might not seem like much — but most of these higher-leverage pitches come in late innings when most pitchers have lost a couple mph off the fastball. Somehow they are able to dial it up and get that speed back and then some. Justin Verlander threw his fastball more than 2 mph faster in high-leverage at-bats than when the game was not on the line. Ted Lilly, Aaron Harang and Pedro Martinez, among others, threw it more than a full 1 mph faster.

It makes perfect sense: When the game is not close or there are no runners on, a pitcher’s best stuff is not necessary, but when the game is close, it’s time to shift to another gear. These higher-leverage pitches almost certainly take more out of a pitcher than when he is cruising.

The Conclusion

Raw pitch counts do not account for the stress a pitcher has experienced over the course of either a game or a season. It is important to track high-leverage pitches separately since pitches in those at-bats require more effort. Here are the 2009 leaders for the number of pitches thrown in high-leverage at-bats:

These guys threw the most stressful pitches in the game in 2009.

Pitcher		Total Pitches	Total High-Stress Pitches
Justin Verlander	3,937	408
Chad Billingsley	3,203	385
Felix Hernandez		3,632	337
Ubaldo Jimenez		3,570	331
Adam Wainwright		3,614	331
Javier Vazquez		3,315	296
Carlos Zambrano		2,843	276
Jon Garland		3,255	271
Barry Zito		3,204	268
Matt Garza		3,421	261

Verlander threw more pitches in 2009 than any other pitcher — and threw the most high-stress pitches. Interestingly, he also was the pitcher with the greatest increase in fastball velocity when the game got tight, which suggests he really worked hard to get out of those situations. A bit worrisome is Chad Billingsley, who ranked only 33rd in total pitches in 2009, yet threw more high-leverage pitches than any pitcher besides Verlander. He might have very well put more strain on his arm than his raw pitch count would suggest.

On the other end of the spectrum are workhorses Cliff Lee and Zack Greinke, ranked sixth and seventh in pitches thrown, respectively, but just 42nd and 35th on the high-stress leaderboard. They likely put less strain on their arms than you might conclude by just looking at their total pitch counts.