|Mariano Rivera has been the king of multiple innings saves in the last decade. Yankees at Orioles May 28, 2008 (Mark Goldman/Icon/SMI)|
Here is a familiar situation: Your team is up by a run in the seventh inning and the manager goes to the bullpen for the seventh inning/middle reliever. You cringe a little, but after a strikeout, pop-up and easy ground ball, it’s on to the eighth on nine pitches with the lead intact. You wonder if the manager might let the same guy go back out, but, invariably, the setup man is called upon.
If that move backfires, and the manager is asked about it later, he probably will mutter something about bullpen roles or relievers not built to go multiple innings or not wanting the reliever to cool down in the dugout and then have to crank it up again.
These are mostly valid concerns. Some old-school guys might say that back in the day Goose Gossage was routinely asked to go multiple innings to preserve a game and we are babying our pitchers today. While it is true about Gossage and others of his day, when they entered the game in the seventh they knew going in that they were there to finish the game and could to pace themselves.
Late inning relievers now are asked to put everything they have into every pitch and generally come out of the game after they have thrown an inning or less. The real question is how a current reliever’s stuff is affected if he has to sit while his team bats. Is the manager in our example correct to go automatically with the setup man, or could the seventh-inning guy pitch the eighth if he worked quickly in the seventh?
For this study I am going to look at relievers who threw at least two innings while the PITCHf/x cameras were watching. I am then going to compare the second inning of work to the first one, using a weighted mean for different PITCHf/x variables like speed and movement. I want to remove long relievers from my data set because when they enter the game they know they are likely to be pitching multiple innings and can pace themselves, so I am going to use only relievers who came into the game in the seventh inning or later with the game within two runs. This late and close situation likely produced a high enough leverage situation that our relievers were giving it their all from the first.
Although I am using the word “inning” here, if a pitcher came into the game in the eighth with two out and got the last out before going back out for the ninth, I am counting that. The reason is I really care about looking at how the downtime affects relievers, so these situations are included as long as the pitcher threw at least two pitches in each inning. I require this because my weighted mean needs at least two objects for each sample to work properly. These conditions were met 429 times from the beginning of this season to July 11.
What I found
Here are the results. A positive horizontal movement implies the ball is moving in to a similarly handed batter.
Fastball Variable First inning Second inning Speed (mph) 91.2 91.1 Horz. move (in) 6.6 6.7 Vert. move (in) 8.3 8.2 Sinker Variable First inning Second inning Speed (mph) 89.8 89.8 Horz. move (in) 8.9 8.3 Vert. move (in) 5.8 5.2 Cutter Variable First inning Second inning Speed (mph) 89.0 89.4 Horz. move (in) 1.8 2.3 Vert. move (in) 8.4 7.5 Curve Variable First inning Second inning Speed (mph) 77.0 77.2 Horz. move (in) 5.8 5.6 Vert. move (in) -4.3 -4.4 Slider Variable First inning Second inning Speed (mph) 83.8 83.8 Horz. move (in) 2.8 2.9 Vert. move (in) 3.0 3.0 Change-up Variable First inning Second inning Speed (mph) 81.7 81.8 Horz. move (in) 7.0 7.0 Vert. move (in) 5.7 5.6
The fastball shows no signs of cracking when the reliever is forced to sit down and cool off for a half inning. The sinker actually produces a little more sink though a little less horizontal movement. This fits in nicely with the idea that sinkerballers generate more sink when they are a bit tired.
The cut fastball is a bit strange. While the velocity is up, so is the horizontal movement. Ideally, the cut fastball moves away from a similarly handed batter, so an increase in horizontal movement is probably not a good thing. The vertical movement on the cut fastball also drops, but it is unclear if that is good or bad for the pitcher. All the off-speed pitches remain intact and, in particular, the change-up keeps its speed differential from the fastball, which we have seen is important.
Okay, so it doesn’t appear that the rest is hurting the reliever’s stuff at all. What if he is forced to throw a lot of pitches in his first inning? Here I required that the pitcher throw at least 20 pitches in his first inning. Because of smaller statistics, here are the results for just fastballs.
Fastballs when the first inning is long Variable First inning Second inning Speed (mph) 91.5 90.4 Horz. move (in) 6.7 6.5 Vert. move (in) 8.5 8.0
Now we see a clear decrease. The speed drops more than a mile per hour and the movement decreases as well. You might not think this is a huge issue, but this is a larger drop than we saw with starters after they had thrown 100 pitches.
Also notice that in this sample the speed started out a bit higher than the total sample. This is likely due to power-type pitchers missing more bats and having to throw more pitches to complete an inning. While missing bats is a good thing for a pitcher, it does drive up his pitch count. Once he has to throw a large number of pitches in his first inning, he isn’t likely to be very useful in the second inning of work. If you reduce the required number to 15, you can see a similar result for curve balls. In the first inning, they drop 4.5 inches but in the second they drop only 3.8 inches.
As long as a reliever doesn’t work an extended inning, it doesn’t appear that his stuff will suffer from cooling off in the dugout for half an inning. Assuming that the current reliever matches up well with the upcoming batters, managers should consider using even their short men multiple innings. If they are sinker/slider pitchers, the results might be even better after an inning of work. If they have thrown an extended inning, however, managers should seek further help from the bullpen to close out the game.