Have We Hit ‘Peak Reliever?’

marc w · May 15, 2017 at 4:42 pm · Filed Under Mariners 

The M’s lost yesterday, painfully, on a Kevin Pillar walk-off HR off of an Edwin Diaz slider. Diaz’s FIP is now 3 runs higher than it was last year, and around 2.8 runs higher than his pre-season projections. The M’s bats were awful in the four-game sweep in Toronto, but as Bob Dutton notes, they’d have won a game or two – despite the offensive woes – with better relief pitching. That’s the context in which I read this article at BaseballProspects about the potentially historic performances (in a bad way) by several bullpens this year. The M’s bullpen has an ERA of 5.22 and a FIP of 4.75, and they’re just dishonorable mentions when discussing the worst ‘pens of 2017. What’s going on here?

Two trends in baseball are more reliable: over time, MLB bullpens pitch more innings and strike out more batters. In 10 years between 2007 and 2016, reliever K/9 went up every single year, and is now essentially at 9, or a K per inning. As the *league average.* Those raised on baseball in the 1980s will never fully be able to comprehend that statistic, but we’ve eased our way to it with small gains every year. At the same time, reliever innings pitched has gone up too, albeit in a slightly different fashion. Reliever’s share of total innings pitched dipped a bit in 2010 and 2011, as the drop in run scoring made starters look better, but it’s risen every year since then. By 2015-16, reliever share of total innings had blown past where it was in 2007-8, and it doesn’t show signs of stopping now.

This makes some sense, given the shift in recent years to building super-pens with two or three dominant set-up men, often highly compensated. This is what I talked about before the 2016 season, when the Yankees assembled Champman/Betances/Miller, and the Royals were coming off a WS win headlined by their pen (Davis/Holland/Herrera). However, despite the big increase in valuation and yet another increase in workload, relievers, as a group, didn’t pitch any better. After bottoming out in 2014, reliever ERA rose in 2015 and 2016, and is higher still thus far in 2017 – same with FIP. The share of inherited runners that score has increased in each year since 2014, when relievers let only 28% of runners score. It was 30% last year and 31% this year, a figure not seen since 2010. Relievers are giving up more runs themselves, and giving up more that are charged to others. That’s not ideal.

It could be a consequence of the growth in workload, though. With more innings to go around, maybe teams are giving more appearances to guys who simply aren’t MLB quality. If *that* were true, we’d probably see each team’s context stats stay the same or improve while the overall run totals and averages declined. It’s one thing to give the last man in the bullpen some garbage time when leverage is lowest, but it’d not like those guys are poaching closers’ innings. This is much tougher to show, as stats like RE24 and RE Wins are much, much more volatile. However, it’s worth noting that bullpens have averaged 269 runs above average in bullpen RE24, or about 29 wins per year from 2007-2016. This year, they’re on pace for 94 runs and 8 wins, respectively. Their context stats have fallen through the floor, implying that this reliever swoon we’ve seen extends far past the Overtons of the world.

As Rob Mains wrote in that BP piece I linked above, he noted that bullpen ERA’s advantage over starter ERA is down this year, and while traditionally a bit less than 1/4 of teams have a bullpen ERA higher than their starters’, 12 out of 30 do this year. I’ve done some of my own digging, and can say that relievers’ share of total runs allowed is higher this year than it’s been since 2007, and presumably a lot longer ago than that. Same with their share of total HRs-allowed. That may be due to increasing usage, of course, so I tried to correct for that by taking the ratio of the pen’s share of runs allowed to their share of IP. When we do, we see that relievers’ share of runs allowed, controlling (poorly, probably) for IP has generally ticked up since 2012.

Share of Runs (Cont.)  
2012 91.6%  
2013 92.9%  
2014 95.6%  
2015 95.2%  
2016 94.1%  
2017 97.7%

By ERA, RE24, FIP, whatever, that 2012 year stands out as a local maximum for bullpen performance. Yes, overall run scoring was lower then, but even still, relievers were dominant vis a vis starters. The gap between relief and starting ERA peaked that year, and the share of runs attributed to the bullpen – accounting for IP – was the lowest in our 2007-2017 sample. Since that time, K rates have gone up, and we keep talking about giving even more innings to the most dominant pitchers of their era (Miller, Betances, etc.), and yet the group keeps slipping. It’s possible that this is a fluke; WPA and context metrics are volatile, after all, and FIP likes the 2017 bullpen crew a lot better than ERA and RE24. Further, the delta between starters’ OPS-against and relievers is unchanged (and higher than it was a few years back). But what if there’s signal hidden in the noise of the context stats? Given that ERA has been rising along with the share of runs allowed, what if hitters are simply better at hitting relievers, even as they’re still more likely to strike out? What if it’s harder to sustain excellent relief pitching performance, and thus relievers really are more volatile than we thought – look at Edwin Diaz, or Darren O’Day, or Sam Dyson, etc? If it *is,* then baseball looks a bit different. It was just a few years ago that we were worried that we’d see fewer and fewer lead changes late in games. That doesn’t seem to be happening.

Ultimately, this may be more of an issue for front offices and how they assign a dollar value to relievers than it is about how teams should utilize their players. The times-through-the-order penalty hasn’t gone away, and it’s madness to suggest the M’s stick with :checks notepad: Christian…Bergman? through the 7th inning rather than handing the ball to the pen and, yes, Diaz. But a year or two after thinking that relievers were still undervalued and that the future of baseball would look more like the 2016 playoffs, I wonder if we might have hit Peak Reliever. The more batters react to and adjust to guys who pop up with dominant reliever seasons, the harder it is to sustain that level of success (even accounting for regression), as you can see by looking at last year’s fWAR leaderboard. Addison Reed, Seung-Hwan Oh, Jeurys Familia and Kyle Barraclough have been replaced by the likes of Corey Knebel, Tommy Kahnle and Ross Stripling. Sure, your name brand guys like Miller, Kimbrel and Jansen remain, but there’s a lot of churn outside that group. Managers need to figure out – quickly – if this is bad luck or batters “booking” certain relievers. Maybe they’re becoming more used to elite velocity thanks to seeing the likes of Noah Syndergaard and James Paxton as starters. Maybe the fact that relievers throw more four-seam fastballs matters, or maybe it’s pure coincidence. I don’t know. But it’s not just Mariner fans who think that the bullpen’s been worse this year, and the implications of that could be far-reaching.


Leave a Reply

You must be logged in to post a comment.