(The following is being syndicated from The Captain’s Blog).
On Saturday, I posted a follow-up to a recent Fangraphs’ analysis of relief pitchers’ aggregate performance over the last 30 years. Although my findings supported the statistical conclusion of the Fangraphs’ piece (i.e., reliever performance has not changed meaningfully over the period considered), there was a divergence with regard to the implications. However, because both analyses only looked at the question from the perspective of relief pitchers, each conclusion may have been incomplete .
Prompted by a comment from MikeD, the following analysis examines the question of bullpen usage from the perspective of the starting pitcher. After all, relievers are only needed to the extent that starters are unable to complete games. Therefore, the usage pattern for the rotation must have an impact on how the bullpen is employed.
Percentage of Batter’s Faced by Relievers and Starters, Since 1982
Source: fangraphs.com
Based on innings pitched and batters faced, it appears as if starters are currently going as deep into games as they were in 1982. However, looking at the aggregate total disguises what may be more meaningful trends. For example, since 1919, there has been a gradual decline in the rate of complete games, and even when compared to 1982, the drop has been significant. But, how does this impact bullpen usage?
Rate of Complete Games, Since 1919
Source: Baseball-reference.com
Going to the extreme, in 1919, starters finished over 60% of their games, so well over half the time, the manager didn’t have to worry much about the bullpen. Fast forwarding to 1982 (one of the parameters of the previously cited analysis), the complete game rate had fallen to 17.4%. Despite a much lower percentage, a manager in that era could still bank on full rest for his relievers coming once every five games. In today’s environment, however, a night off for the bullpen is a very rare occurrence. As a result, modern managers are faced with the challenge of spreading their bullpen assets over a greater number of games.
If complete games have decreased significantly, what explains the relatively stagnant workloads indicated by the Batters Faced chart presented above? Could it be that the demise of complete games is merely anecdotal, especially if they are being curtailed simply out of deference to the save rule?
Rate of Innings Per Start, Since 1919
Source: Baseball-reference.com
Looked at more broadly, the rate of pitchers going beyond six innings has declined from over 75% in 1919 to 54% in 1982 and then 43% in 2011. Meanwhile, the number of starts lasting three innings or fewer has also declined, falling from just over 10% in 1919 and 1982 to under 5% in 2011. Because of this convergence, there were more starts that lasted between five-plus and seven innings in 2011 than at any other point since 1919.
Rate of Starts Lasting Between 5 1/3 and 7 Innings, Since 1919
Note: Excludes appearances in which the pitcher appeared in the sixth or seventh inning, but did not record an out.
Source: Baseball-reference.com
Considering the patterns presented in the charts above, it seems as if the evolution toward shorter (but more frequent) appearances has taken place out of necessity. Because very few starters progress into the later innings, managers now must use their bullpens extensively in almost every game. As a result, a “lose the battle to win the war” mentality has taken hold. Whereas 30 years ago, a manager could use a top reliever for several innings because he could count on one of his starters throwing a complete game in the near future, today’s skippers aren’t afforded that luxury. What’s more, it seems as if the need to spread out reliever appearances has resulted in managers being less likely to turn the game over to the bullpen in the early stages. Without room on the roster for a true long guy who could be exhausted for an extended period of time, “saving the bullpen” has become another common mantra for the modern manager.
One final point worth noting is just because relievers aren’t throwing more innings or facing more batters doesn’t mean they aren’t working harder. Along with more appearances come more warm-up throws (what managers today call “getting hot”), many of which occur in games the reliever never enters. Although this doesn’t show up on the stat sheet, it is something that managers take into account when making decisions about their relievers.
Whether its “lose the battle” or “save the bullpen”, the modern approach to managing a bullpen seems designed to mitigate the risks inherent in having a starting rotation that is unable to provide length on a regular basis. Based on the evidence, it doesn’t seem as if managers could simply reverse course and readopt the bullpen management philosophies of the past. Instead, innovations in bullpen management should be focused upon optimizing leverage, or other modifications to current strategy, because today’s starting pitchers simply wont allow relievers to be used as they were in bygone eras.
If you enjoyed this article, please consider sharing it!
7 Responses to Starting Over: Looking at Bullpen Usage from the Other End
Leave a Reply
TIQ IQ
Of course they could! Not immediately of course since starters are now conditioned to shorter starts. But it doesn’t seem to me and I’d be surprised if it were supported by evidence, that this has made pitchers any more effective or less injury prone. All it’s done is transfer innings to less effective pitchers (middle relievers) and raised their risk of injury since they now work more often.
You might argue that the changes in starter and reliever usage patterns are supported by evidence. However, you can’t argue that these changes were initiated by a data driven process. They were initiated, particularly in the case of relief pitcher usage patterns, by fad and the distortions created by the save rule. Post-hoc rationalizations of why it has to be this way aren’t too convincing.
I’m not arguing for a return to the days when starters were throwing 20-30 complete games per year (although I wouldn’t mind seeing more CGs). What really galls me are 1) the proliferation of 12-man pitching staffs bloated by the addition of fungible middle relievers whose importance is exaggerated beyond their actual ability 2) LOOGYs (I can’t really say anymore without going all Lewis Black) 3) the “slave-to-the-save” strategy that dictates the use of the most effective reliever, often preventing him from being used where he could do the most good and 4) the so-called “structured bullpen” where guys are designated to pitch a certain inning irrespective of leverage and get bent out of shape or feel dissed if they’re asked to pitch in a different situation.
Look, having to rely on 3-4 pitchers per game is nothing less than an algorithm for finding a guy who’s having a bad enough night to cost you the game. Since starters are typically better than middle relievers you’re more likely to find that guy in your bullpen.
The save rule has been around a long time, so I am not sure why you think it was the main catalyst for the modern bullpen approach. Rather, I think shorter appearances evolved over time and the save rule just became a convenient rule of thumb that was further ingrained by the prominence of the statistic. While I agree with you that managers and players should try to wean themselves off strict adherence to the save, I don’t think that means they can go back to an era in which there were fewer, but longer appearances, for all the reasons I cited above.
I also don’t think starters can gradually be conditioned to throw more innings unless the pitch burden also subsides. Unfortunately, there isn’t extensive pitch count databases, so we can’t make meaningful comparisons, but I suspect on a per pitch basis, starters are not slacking as much as many seem to believe.
As for the injury aspect, I can’t speak to that because I am not sure how you’d prove causation.
Yes, the save rule was around for a while it became the holy grail but there’s no question that it had a significant influence on reliever usage patterns. Otherwise how can you explain the way closers have been used since the eighties? I also didn’t say it was the only influence. In any case, I really doubt there was a lot (or any) number crunching behind how these usage patterns evolved.
With all due respect I’m not impressed by arguments how the way things are today are inevitably the way they need to be. After all, they didn’t start out that way and while changes will, of necessity have to be introduced gradually, I find it difficult to believe that someone won’t get a better idea (we hope based on actual data – but probably a marriage of that and circumstances) and implement it.
I don’t think there was any number crunching behind the evolution toward shorter appearances. Rather I think it was necessitated by starters individually failing to go deep into games.
I am not sure where you’ve came away with the impression that I think the status quo is unchangeable. On the contrary, my conclusion is that although shorter appearances are necessitated by fewer long starts, adjustments within the framework might enhance reliever value. However, that doesn’t mean the best approach is returning to usage patterns 30 years ago. If you disagree, then how do you propose to increase the innings that starters throw? If you are going to ague in favor of longer relief appearances, you must answer that question.
I’d be curious to see how pitch counts have changed (if at all) over time while innings have dropped. I would wonder if the more recent emphasis on OBP as a measure of value may lead hitters to take more pitches, thereby shortening the number of innings pitchers can throw. There’s probably no connection, but it’s one theory.
Well I don’t actually have data (or time to collect it – even if it were available) but I have been watching a lot of baseball for a lot of years (back to the sixties) and it’s not my impression that the changes we’ve seen in starter and reliever usage patterns were dictated primarily by an intrinsic inability of starting pitchers to go more than 5 or 6 innings. Sure, there are some that can’t (and there always were) but I think it’s more the case that this usage pattern somehow became the prime directive for all starters irrespective of whether they could go further or not. In other words, teams adopted this pre-emptively on a one-sized fits all (OK, most) basis.
Back in the day, pitchers knew how to pace themselves and how to save pitches and effort on the lesser hitters in the lineup so they could throw more innings. OK, lineups (especially in the AL) are deeper and stronger so there’s less coasting that you can do. I get that. But it seems that pitchers are encouraged not to even try to go further but to “empty the tank” go all out for 5 or 6 and then the pen will bail you out.
That’s the difference and that’s why I emphatically disagree with your stance that pitchers could not be conditioned to go longer in their starts. Of course they could since it’s not a matter of intrinsic ability but a strategy that’s involved. Well, in my opinion the strategy is a poor one and furthermore it necessitates the change in reliever patterns that we’ve seen come about for which statistical analysis has failed to show a benefit.
As you acknowledged, a lot has changed since the 1960s. Since that time, which was favored pitchers significantly, just about every adaptation until recently tilted the balance towards offense. Unfortunately, we don’t have pitch data from the 1960s, but I’d wager that it took many fewer pitches to complete a game in that era. You simply cannot dismiss.