A Rebuttal to Bullpenning
The term “bullpenning” has recently risen to prominence in the baseball lexicon, referring to the practice of deploying relief pitchers earlier in the game to exploit matchups and avoid the dreaded times-through-the-order penalty that afflicts starting pitchers. The movement has been ongoing for years, and last year’s playoffs pushed it into the foreground of baseball strategy. Fans are being told to expect records amounts of it in the upcoming playoffs.
The argument is that starting pitchers fare significantly worse when they go through a lineup more than twice, and replacing those tiring pitchers with fresh relievers creates an advantage for the team in the field. In a single game or short series, the potential benefits are obvious.
But professional baseball is a game of adjustments; hitters will react to any strategy over time.
In particular, I’m interested in how this strategy plays out over the course of a long playoff series. Just as the 2016 Indians serve as the shining example of a team bullpenning its way to victory, the 2016 Cubs may have provided the blueprint to combat this strategy.
As a Cubs fan, one of the more rewarding aspects of last year’s playoff run was watching them adjust throughout each series. Casual observers may have watched the first few games of each round and observed that their bats were overmatched by the plethora of bullpen arms unleashed against them. But if you watched closely enough, you could see the purpose in each at bat, and then the progress in each subsequent at bat, ultimately paying off in the form of key hits against the same relievers who had previously shut them down.
In other words, major league hitters are very good at what they do, and they learn from each at bat. When they face the same pitcher, starter or reliever, within the short timeframe of a playoff series, they will make adjustments in order to improve their outcomes. Anecdotally, the Cubs exemplified this in 2016 (or maybe that’s just my biased optimism talking again), and I would expect to see similar improvements by other teams, as well. Let’s see if the data support my theory.
Using the Baseball Savant Statcast Search tool, I collected data on all 2,587 at bats from the 2016 playoffs. Then, I grouped the data by pitcher/hitter matchup and sorted by game date and at bat number, looking to see how batters fared in subsequent at bats against the same pitcher. For simplicity, I’m going to refer to the “time-through-the-order” as the matchup count for each pitcher/hitter combination. The chart below shows how WOBA values changed as the matchup count increased, for both starters and relievers. The dotted line represents an average WOBA, according to FanGraphs.
Based on WOBA, the results for starters and relievers were comparable during their first, second, and third times facing the same batters. However, hitters feasted on relievers when seeing them a fourth or fifth time in the same series (though you should take that with a grain of salt due to very small sample sizes). The table below offers a closer look at a few metrics, and includes the number of at bats for each matchup count. The green highlighting indicates which group of pitchers fared better.
Relief pitchers outperformed starters the first or second time they were used in a series. It gets more complicated the third time through and beyond. Obviously, there are other factors involved, but the table shows that starting pitchers fared much better than relievers on their fourth and fifth trips through a lineup.
There’s a problem with this table, though. It doesn’t paint an accurate picture of a manager’s dilemma in these scenarios. So let’s assume that the starter was able to get through the lineup twice, and look at the numbers that way.
The first few columns support the idea that fresh relievers fare better than tiring starters. However, the results shift back in favor of the starters as you move further to the right. Relief pitchers on their third time through the order performed better than starters on their fifth, but not by much. Relief pitchers on their fourth trip through looked better than starters on their sixth, but that changes if we compare them to starters on their third or fourth time through the order. It’s a lot to think about.
For example, pretend you’re the manager of a playoff team during the middle part of game 5 or 6. Your starter, possibly your ace, is facing the same hitters for a fifth time. You have narrowed down the bullpen to only those you trust, so everyone has pitched at least once at this point in the series. The table to the right suggests that you may be faced with a very difficult decision.
Again, these are small samples, from only a single postseason. Also, there is some natural bias here, as well. The starting pitchers who face batters for a fourth, fifth, or sixth time in a series are typically their team’s ace, or at least their #2, so that could skew the data towards the starters. Also, I identified each pitcher as either a SP or RP, so the “starters” group received credit for any relief innings thrown by a pitcher who started a game at some point in the playoffs. There is certainly more work to do on this topic, including expanding it to see how it holds up in previous postseasons, so I will try to refine some things along the way.
At the very least, this suggests that managers should exercise caution before going all-in on bullpenning this postseason. Yes, it’s better to be ahead in a series early than behind; but teams that rely on their bullpen early and often in this year’s playoffs may ultimately find themselves unable to finish off their opponents without an adjustment of their own.
Check out the code here