A Reliever Usage Check-In
True to form, 2020 has been a relief-heavy endeavor. As Jay Jaffe noted, starters are on pace to throw their fewest innings per start ever. You can do the math — that means that relievers are on pace to throw their most innings per game ever. Given all that, here’s a question for you: what does that mean for reliever rest and usage patterns?
First things first: I used one broad cut to separate relievers in two groups. I looked at the 90 best projected relievers by ERA for both 2019 and 2020, which produced a cutoff of a 3.945 ERA in the 2019 projections and 4.01 in the 2020 projections. This isn’t a perfect way of separating the wheat from the chaff, but it importantly holds both sample populations the same.
In 2019, those 90 top relievers accounted for 19.5% of all batters that relievers faced. There were 683 relievers who pitched at least one inning last year — and this small group accounted for a lot of it. Has their usage share been diluted with a new group of relievers in 2020? Not at all — in fact, they’ve accounted for 23.7% of relief batters faced this year, despite the larger benches.
There’s a slight sampling bias inherent in this calculation. We’ve only seen three weeks of baseball in 2020, which means less time for teams to change their opinions on which of their relievers are best. By September 2019, a preseason projection of a 3.50 ERA meant less than it did in March. Still, I’m not ready to dismiss the finding. Teams want their best relievers on the mound more often when they feel like every game counts, and they’re absolutely in that mindset at the moment.
Another characteristic of elite relievers is that teams want them on the mound in the biggest spots. To wit: the average reliever in 2019 entered the game in a relatively unimportant spot. Leverage Index has its downsides, but it’s great for noting how big of a potential swing a plate appearance can produce. Man on third and one out in a tie game? That’s a high-leverage spot. Up five with three outs to get? Send on Johnny Mop-up. Relievers, as a whole, entered the game with a 0.92 LI on average in 2019, where 1.0 is neutral.
Good relievers, on the other hand, came in when the game was close. They averaged a 1.08 entry LI in 2019. That’s not rocket science, but it’s a useful data point. I didn’t have an obvious theory either way as to what would happen in 2020, so I simply ran the numbers. The best relievers in 2020 have entered the game with an average Leverage index of 1.16. The general principle isn’t surprising: teams bring in their best relievers when they think they can do the most good. The same sampling bias that affected innings pitched might affect entry LI; it’s hard to say if the best relievers are being used in bigger spots in 2020 than in 2019.
Though the top-line numbers are similar, averages lie. Think of relief appearances in two buckets: appearances after a day of rest, and appearances when the pitcher appeared in a game the previous day. Rested and elite relievers came into games in 2019 with an average LI of 1.04. Teams used them in situations that were more important than average, but not by a ton. In 2020, that pattern has continued; the average entry LI for rested elite relievers has been 1.08.
What would cause a team to bring back a good reliever on no rest? Why, important situations, of course. In 2019, teams brought their top relievers in on no days of rest with an average entry LI of 1.25. That’s an abstract number, but some relievers who entered with a Leverage Index in that range last year were Nick Anderson, Jose Leclerc, and Archie Bradley. If one of these top relievers came in on back-to-back days last year, it was usually for good reason.
This year, that effect has gone into overdrive. When top relievers have appeared on zero days rest, it’s been with an entry Leverage Index of 1.5. That’s stopper/closer territory; Liam Hendriks, Alex Colome, and Ryan Pressly were in that general range in 2019. It’s early to say this for certain, but it appears that teams are far more judicious with their back-to-back pitching days for their best relievers this year.
Another supporting data point: in 2019, 20.5% of appearances by this top group of relievers came on zero days rest. They faced 3.8 batters on average in those appearances. This year, that share of appearances on no rest has fallen to 17.6%, while average batters faced has risen to 3.9. It seems teams are being more judicious with the second day of pitching — which feels reasonable when bullpens run to nine and 10 pitchers instead of eight.
My theory here is that bullpens are less prone to depletion when they have extra arms, so low-leverage uses of good relievers have become rare. In 2019, 30% of no-rest appearances by great relievers came with a Leverage Index higher than 1.75, true fire-alarm moments. That number is up to 36% this year. Meanwhile, 2019 saw low-leverage back-to-backs (with an entry LI below 0.8) 38% of the time. This year, that’s down to 25%.
Why use a good reliever in a low-leverage spot when he pitched the previous day? Plenty of times, there’s simply a total supply constraint. If a team needs five innings of relief work on two straight days, at least two relievers likely need to pitch back-to-back days in a world of eight-man bullpens. Why not use someone good if you’re going to double up somewhere? This year, there’s less need for that.
Indeed, overall appearances on zero days rest are down. Last year, there were 2,977 total relief appearances on no rest, a rate of 0.62 per game. This year, that rate has fallen to 0.48 per game; in a full season, that would be 625 fewer such appearances. Bigger bullpens have led to more even rest schedules, even with the increased bullpen workload due to shorter starts.
Rest patterns and top reliever usage rate are the two main changes I’ve noticed this year. Aside from that, I haven’t been able to find much of a pattern. In 2019, good relievers faced 4.1 batters per appearance on average. In 2020, that number is still 4.1. That’s more of a population issue than a strategy issue — it isn’t easy to suddenly make your one-inning relievers into two-inning relievers — but it’s useful to know that nothing has changed there.
Pitches per appearance are also static; 16.5 per game for our top group in 2019 and 16.6 per game in 2020. When the good relievers are in, managers appear to be sticking to their old guidelines for how long to leave them in. Batters faced and pitches thrown on zero days rest are also unchanged.
Before we hit the conclusion of the article, here’s a quick table to summarize top reliever usage in 2019 and 2020:
| Metric | 2019 | 2020 |
|---|---|---|
| Share of Relief TBF | 19.5% | 23.7% |
| Entry LI | 1.08 | 1.16 |
| 0 Day Rest % | 20.5% | 17.6% |
| LI on 0 Days Rest | 1.25 | 1.5 |
| 0 Days Rest, Low Lev % | 38% | 25% |
| 0 Days Rest, High Lev % | 30% | 36% |
| BF/Appearance | 4.1 | 4.1 |
As starters build up their stamina and bullpens shrink while rosters go from 30 to 28, some of these usage patterns will change. Teams will face a numbers crunch more often, but they’ll also presumably have fewer innings to cover. Additionally, as playoff races heat up, teams might be more willing to push their best relievers in big spots.
When either of those happens, we’ll be here to chronicle it. For now, all I can say is that teams seem to be behaving rationally with their extra relievers. They aren’t getting fewer innings out of their best arms, but they are spacing those innings out more; less back-to-backs, and more important appearances. In a year that’s been hard to predict in almost every other way, this one seems to be exactly by the book.
Ben is a contributor to FanGraphs. A lifelong Cardinals fan, he got his start writing for Viva El Birdos. He can be found on Twitter @_Ben_Clemens.