Notice: Trying to get property 'display_name' of non-object in /var/www/html/wp-content/plugins/wordpress-seo/src/generators/schema/article.php on line 52
keyboard_arrow_uptop

The complete game has become an increasingly rare beast. In 2013, there were 124 complete games registered by the 4,862 pitchers who started out on the hill, and Adam Wainwright led all of baseball with five. If a pitcher makes it through nine innings, he’s likely having a very good day, and nine innings of well-pitched baseball is nothing to sneeze at. But a complete game is more than that. It’s a sign of manliness. It’s like shouting, “I don’t need no stinkin’ bullpen!” It’s a cultural touchstone. It’s the guy yelling at his TV, “Finish what you started, you silly overpaid, coddled millionaire. I finish my day of work without calling in a reliever.” A pitcher who completes a game is just getting in touch with the common man.

There’s a certain fascination surrounding the pitcher who goes deep into games, even if he doesn’t complete them. It’s been argued that such a player provides value well beyond the confines of the game in which he pitches. Logically, the argument makes sense. Coin Wyers pointed out that the deeper into a game a pitcher goes, the better the bullpen performs. Mostly, that has to do with the fact that if a manager can get through the seventh with his starter, he can tell his third- and fourth-best relievers to take the night off. In some bullpens, the last thing you want is that fourth-best reliever throwing any pitches. Others have suggested that if a manager has a starter on whom he can count to go deep into a game, it affects his choices on how to work the bullpen the day before (he doesn’t have to “save” as many arms for the next day) and after (all of the relievers are rested). In that way, the complete game affects three different games!

And then someone says it. “How do you really figure out the full value of the complete game? I mean, there’s the game itself, but there’s all that value that he’s adding in the game on either side, and that’s not being captured in the stats. There’s just no way to fully value that.”

I do believe that was the sound of a gauntlet being thrown down.

Warning! Gory Mathematical Details Ahead!
To solve this problem I decided to create the perfect pitcher. He is so good that his manager knows that on his day, no one in the bullpen will need to lift a finger. They might as well not be at the ballpark. It’s 100 percent guaranteed. On the day before his scheduled start, the manager can use the bullpen to his heart’s content. On the day after, he’ll have a rested pen.

The funny thing is that this pitcher already exists, and no one gives him any credit. Ladies and gentlemen, I give you the perfect bullpen saver: Scheduled Offday.

Every season, a manager has a handful of situations in which he knows for sure that his bullpen will not be needed tomorrow. Worried about getting four outs from your closer because it might mean that he won’t be available tomorrow? Don’t worry. There is no tomorrow!

Thanks to Retrosheet, we have files that show the original schedule for a season. (Of course, rainouts and other unfortunate events change what actually happens, but that’s life.) Using these files, I looked for games where a team had an offday scheduled for the next day. To make sure that no makeup games were snuck onto the schedule, I made sure that no game was actually played the next day as well. I looked to see how managers behaved on the day before an off-day versus when they did not. All numbers are from 2009-2013.

So let’s see how well each of the standard arguments holds up:

The manager can have a quicker hook with his starter, because he can extend his bullpen a little bit.

Before a day off

No day off

Batters faced by starter

25.09

25.26

Outs recorded by starter

17.50

17.59

It’s a tribute to the sample sizes that we’re dealing with (24,000 games!) that the difference between the numbers of batters faced is almost significant (p = .059), but that’s not even a one percent change over baseline.

The manager can use more relievers and mix and match more effectively. He can feel more free to use a pitcher for a batter or two.

Before a day off

No day off

Number of pitchers used (inc. starter)

4.04

3.90

Number of pitchers who faced 2 or fewer batters

0.53

0.46

Percentage of games featuring a pitcher who faced 2 or fewer batters

39.8%

34.6%

Here we see that before a day off, managers do feel at liberty to make more calls to the bullpen, although again, the effect sizes are not huge.

The manager can use all of his good relievers, rather than holding one back in case he should be needed tomorrow. He can also extend his good relievers a bit.

Before a day off

No day off

Average Reliever RA/9 in game

4.15

4.42

We can see that there is a difference between the types of pitchers used by managers before a day off and those used when there’s a game the next day. At first blush, this might seem like a big deal, but let’s drill down a little further. The difference in results for a day off was 0.27 runs per nine innings. Since we know that managers aren’t going to their bullpens earlier, we can assume that relievers are picking up the same basic workload.

In 2013, the average major league bullpen pitched just shy of 500 innings (499.23, to be exact). We can figure that if managers always used their bullpens the way that they did before a day off, they would save 15 runs (actually a little less, because some of that is already baked into the results, but let’s ignore that for a minute). That’s not insignificant. The strategic ease that comes from the ability to say “tomorrow, I won’t have to worry about the bullpen” is like having a ghost 1.5 win player on the roster. The problem is that the manager has that luxury only a handful of times per year.

To make sure that there weren’t weird things going on, I limited the sample to games that were reasonably close, defined as games that were within three runs either way at the start of the seventh inning. I also took out any games after September 1, figuring that the expanded rosters might play havoc with the findings. The numbers, of course, wiggled a bit, but the overall message was the same.

Now, let’s come back to reality. This would necessitate a situation where a team played games only on alternating days so that the manager absolutely knew that tomorrow was always a total freebie, or that the manager always had totally delusional levels of confidence that tomorrow, his starter would throw a complete game. If we assume that managers are not actually delusional, then we would expect them to understand that while a pitcher might often give him seven or eight strong innings, and could very well throw a complete game tomorrow, he at least needs to consider what tomorrow will bring. There goes some of that ease factor.

On top of that, any one starter is going to be tomorrow’s starting pitcher only about one-fifth of the time. So, even if we assumed that Mike Matheny managed his bullpen differently when Adam Wainwright was scheduled for the next day, that ease is present in only 30-something games per year. The upper limit of an effect that one pitcher could have from a reputation as a complete gamer would be 0.3 wins or so, and that’s if the manager believed that he would always start and finish the next game. The effect is probably smaller, and while 0.1 or 0.2 wins of additional value per year isn’t irrelevant, the order of magnitude is well below what it’s made out to be by those who sing the praises of the complete game.

But wait, what about the next day’s games? If a pitcher throws a complete game, then the bullpen had a day off and the manager has some leeway to use pitchers a little more liberally, right? I identified games that were played after a day off and found the following.

After a day off

No day off

Batters faced by starter

25.16

25.26

Outs recorded by starter

17.40

17.60

Number of pitchers used (inc. starter)

4.01

3.90

Number of pitches who faced 2 or fewer batters

0.50

0.47

Percentage of games featuring a pitcher who faced 2 or fewer batters

36.2%

35.0%

Average Reliever RA/9 in game

4.39

4.43

When a manager has the luxury of giving his bullpen the previous day off, they do pitch a little better, and he’s a little more likely to use them. Whether that’s because he’s able to optimize matchups or because the arms are rested, there is an added effect. If a team was somehow able to have its bullpen pitch all game with its members having always had the previous day off, the effect would be about two runs over the course of a season. Of course, that luxury happens only when the previous day’s starter actually completes a game. Even if a starter completed all 34 of his starts, the benefit for the next day would be just a fraction of a run.

The Cult of Complete
I can appreciate the value of a complete game (or even a good eight-inning outing) unto itself. It’s usually also a well-pitched game, so huzzah for that. And yes, there are effects beyond the game itself. But those effects are relatively minor, and on the scale of maybe a small fraction of a win over the course of a year for any one pitcher. And that’s the upper limit. “How his complete game sets up the rest of the team” is nice to talk about, but once you run the numbers, you see that the actual effect is much less than those who sing its praises seem to believe. Maybe it does make the manager feel better about life, but you can’t put happy feelings on the scoreboard.

If you want to make the case that a pitcher who throws a lot of complete games lessens the workload on the players in the bullpen over the course of a season and keeps their arms fresher and perhaps less injury prone, that’s a reasonable hypothesis (for another day). But let’s not make the complete game or the pitcher who goes deep into a game more than he really is. The knock-on effects that it has are not some deep mystery that takes 50 years of study to fathom. They can be easily estimated. The effect of those heroics is not zero, but let’s keep some sense of perspective.

Thank you for reading

This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.

Subscribe now
You need to be logged in to comment. Login or Subscribe
mgolovcsenko
4/01
Love the approach - using the offday as a simpler proxy for complete games. That's the kind of seeing-beyond-the-obvious problem structuring (& then backing it up with the analytics) that keeps me subscribing.
bhalpern
4/01
I wonder if the difference in reliever usage would be more or less significant when analyzing samples of two consecutive days played before a day off versus similar groups of games not preceding a day off. Also, since we're now in an era where it's already fairly normal to use a reliever for two or less batters almost daily maybe excluding the relatively recent era of this type of reliever usage might show there was a more significant day-off effect in the past.
oldbopper
4/01
When I read "I don't need no stinkin' bullpen" the Mets instantly came to mind. Is at possible to believe that Jose Valverde is the best they have?
kingstriker03
4/01
I know there are pitchers that statistically become more dominant later in games, but I have to assume they're few and far between. Based on the fact that your study seems to show that the advantage of a SP throwing a CG is minimal, it seems likely that a team actually has a statistical advantage in pulling a SP late in the game in favor of a dominant setup guy or closer...regardless of how well the SP is throwing. The odds of that SP breaking down and making a bad pitch due to exhaustion might be higher than a fresh closer coming in to finish off the game. Of all of the CG's thrown in your study, I'd be curious to see the breakdown of what the run differential was entering the 9th inning.
gjhardy
4/01
Very interesting. Since a manager can never assume the SP's performance in tomorrow's game, almost all the value of the CG is tied to the game after the complete game, right? But your chart shows that although more RP's are apparently available to pitch, they don't pitch as well, with 4.43 RA vs. 4.39. Isn't that actually a negative effect?

There might be some unexpected noise in here too: Some managers might, on occasion, use an off-day to skip a fourth or fifth starter's turn in the rotation, or to juggle the rotation a bit for matchup purposes.
pizzacutter
4/01
That should actually be reversed. 4.39 after a day off, 4.43 with no day off. It's a very small effect.
bornyank1
4/01
Fixed.
rweiler
4/01
If any team was lucky enough to have a rotation full of guys that could pitch complete games on a fairly regular basis, they could get by with fewer relief pitchers and thus be able to add an extra bat or two to come off the bench. That might add a win per year. Of course any team that had 5 guys pitching a lot of complete games would probably be winning most of them anyway.
DetroitDale
4/01
Two excellent points, which dovetail with some ideas, I've had. The evolution away from the complete game and towards highly specialized bullpens has resulted in 12-13 roster spots to pitchers where previously 9-10 was sufficient. This results in a much shorter bench, usually a backup catcher, a 4th outfielder and a guy who can play every position but can't hit a lick. For this reason, I've been advocating expanding the roster to somewhere between 28-30 players.
DetroitDale
4/01
Excellent analysis. Surely a bullpen would benefit from a day off (though not as much as we thought by this analysis) but not every team has a Justin Verlander who can give them one. For those teams, I have an alternate suggestion, the 6th starting pitcher. Every team usually has two guys fighting for the fifth spot on the roster and one ends up either converted to a reliever or sent to AAA. I say, keep the odd man out, but run out there every fifth day, the same day as the guy who beat him for the fifth starting spot. One guy pitches innings 1-5, the other 6-9. Each guy only faces the hitters about twice, avoiding the third trip through the order that really causes problems, you can rest the other pitchers that day, and you can save your closer and setup guys for the other days without giving any innings to those middle relief guys that REALLY make you cringe.

garthhewitt
4/01
I believe the perceived greatest value of a CG is the "stopper" value. Ie the effect is most appreciated when the bullpen has been stretched due to poor starts prior to the complete game. The testable hypothesis is something like: When a complete game follows multiple short starts, the team ERA is better in the next four games than in the preceding four games. If there is a stopper value, it may be hidden in the statistical noise when compared to overall averages.
alangreene
4/02
So is there really no real value on other games, or are managers not taking advantage of something they should?

At minimum, it seems like they may be wasting off days.
gjhardy
4/02
I would hate to see MLB expand rosters. Managers and Front Offices should be forced to make some decisions on who to keep on the active roster; with 28-30 players, we would be looking at more relief pitchers, more platoons, etc. The All-Star Game is pretty much unwatchable; I would not like to move in that direction at all.

I think a smart FO would try to find some kind of efficiency with the fifth starter position. I would love to see the numbers for the "average" fifth starter last year. Assuming this guy -- or this collection of guys -- starts 30 games in a season, I suspect that teams are essentially handcuffing themselves in something like 10% of their games by trotting out a really marginal guy in that slot.

That's fine if you are not a contender; you can use the fifth slot to develop a young guy. But if you are in contention in September, you might regret giving away multiple starts in April-May to a guy who you already knew was not good.
harrypav
4/02
"Maybe it does make the manager feel better about life, but you can’t put happy feelings on the scoreboard." Not for long. FEELSf/x is coming soon, and it will be made available to the scoreboard operators.