The Ever-Changing Landscape of Second-Tier Relief Pitchers

Aroldis Chapman, Zach Britton, Mark Melancon, and Jeurys Familia are on the disabled list. Edwin Diaz, Seung Hwan Oh, Kelvin Herrera, and Sam Dyson have had major difficulties. Greg Holland, Justin Wilson, Corey Knebel, and Bud Norris have been unexpected studs.

If you find yourself frustrated with the volatility of relief pitcher performance, you’re probably not alone. The names above represent just a handful of seemingly dozens of unexpected developments in the relief pitcher landscape that take place throughout the course of season and affect major league rosters and fantasy teams alike.

For example, a few weeks ago, I traded $6 Edwin Diaz and $3 Carter Capps for $9 Mitch Haniger in an Ottoneu fantasy league. Later, I flipped the $9 Haniger for a $9 Gary Sanchez.

At first, I assumed that I probably lost the trade by buying so high on Haniger. After all, Diaz was an extremely hot commodity after his scintillating performance as a 22-year-old rookie last season. When other owners and I discussed trades, Diaz was almost always pointed to as a primary target.

Shortly after the trade, however, Diaz more or less fell off a cliff. In his last two outings, he has allowed a walk-off home run to Kevin Pillar, and walked four batters in a row with a three-run lead in the ninth before being pulled from the game.

Overall, in 15.1 innings this season, Diaz’s 29.4% strikeout rate is 11.2% below last year’s mark, and his 71.8% contact rate is 8.3% higher than it was last season. Add it all up and Diaz has an unsightly 5.28 ERA/5.95 FIP/4.42 xFIP, bringing his career numbers to a 3.36 ERA/2.94 FIP/2.46 xFIP, which is by no means bad, but is worse than expected after he put together a 1.88 xFIP last season.

Diaz is expected to be much better for the rest of the season. We are, after all, talking about just 15.1 innings. Steamer projects a 2.99 ERA and 3.10 FIP in 49 innings, and ZiPS forecasts a 3.25 ERA and 2.91 FIP in 58 innings. Both systems expect Diaz to amass more than 11 strikeouts per nine innings. ZiPS projects a robust 13.56 K/9.

The point is, however, that Diaz has lost some value, whether or not that’s fair. And he’s not alone. Relief pitchers throw so few innings from year to year that their results can fluctuate dramatically. As pitchers, they’re also seemingly more susceptibility to serious injury than position players.

Of course, it was Haniger, not Diaz, who got hurt — but that is not the point. Success for relief pitchers seems to be among the most volatile in baseball.

I wanted to test this hypothesis, so I looked at the top 25 relief pitchers in 2015, 2016, and 2017, based on FanGraphs points per inning in the Ottoneu fantasy baseball platform. For all three seasons, only qualified pitchers with zero starts were included.

FanGraphs points are based on linear weights, so the totals do a respectable job of reflecting a pitcher’s true value (unlike most fantasy baseball scoring systems). Saves and holds are rewarded in FanGraphs points. Click here for a full breakdown of how FanGraphs points are calculated.

Getting right to the results, first, here are the top 10 fantasy relief pitchers, sorted by FanGraphs points per inning, for 2015, 2016, and 2017. The three top 10s are immediately followed by some brief analysis, and an introduction to the top 11-25 relief pitchers per season. An explanation as to why the charts have been broken down in this way will be given following the presentation of the data:

The top 10 in 2017:

The top 10 in 2016:

The top 10 in 2015:

A lot of the names in the top 10 across the three seasons are actually the same. This seems to go counter to my original hypothesis that relief pitcher volatility is rampant, and that the top relievers are not reliable from year to year.

But I wasn’t really thinking about just the top 10. I was thinking about the top 25 or so. Check out the names in the top 11-25, and see if you can spot a different trend:

The top 11-25 in 2017:

The top 11-25 in 2016:

The top 11-25 in 2015:

It can be hard to tell with so many names, but an interesting trend emerges from the top 11-25 players, especially when compared to the top 10 players.

For players that do appear in the top 10 in any given year, the mean frequency of those players appearing in the top 25 in all three years is 2.1, meaning that those who cracked the top 10 in any year from 2015-17 are in the top 25 78% of the time across the three seasons. That’s pretty high.

However, among relievers appearing in the top 11-25 in any one of the three seasons, those players have a mean frequency across all three seasons of just 1.53, or 51%. Interestingly, the ratios (51%) are exactly the same for all three individual seasons.

Put another way: If a pitcher has performed as a top 11-25 relief pitcher by FanGraphs points per inning in 2015, 2016, or 2017, he’s only been in the top 25 an average of 1.53 out of three times, or 51% of the time.

While this research is far from statistically significant and doesn’t prove anything, it certainly seems to suggest something about the volatility of the top 25 relief pitchers from year to year.

This season, 10 of the 15 names in the top 11-25 did not appear in the top 25 in 2015 or 2016; in 2015, nine of the 15 names in the top 11-25 do not appear again in the top 25 in 2016 or 2017; in 2016, nine of the 15 names in the top 11-25 only appear once on the three leaderboards.

In most fantasy leagues, the top 11-25 relief pitchers are extremely relevant. If a reliever is in the top 25 in FanGraphs points per inning, he’s almost certainly near-universally-owned in holds leagues, for example.

But since the top 11-25 relief pitchers seem to have such a high turnover rate, it could mean that fantasy owners should not get overly attached to them. While there are a few consistently reliable bullpen arms, there is also a ton of uncertainty. Even some of those “reliable arms,” like Chapman and Melancon, are experiencing health issues this season.

New relief pitchers seem to enter (and depart from) the top 25, especially the top 11-25, on a regular basis. Further research is needed to see if this trend holds over greater periods of time, and if it differs from the volatility of starting pitchers or position players. Those are questions that may be explored in future posts.

Coming back to the Diaz-for-Haniger and Haniger-for-Sanchez trades that set this whole thought process in motion, again, it was Haniger and Sanchez, not Diaz, who got injured. As far as we know, Diaz is healthy, and his sub-par performance this year comes in an extremely small sample. It’s entirely possible, and perhaps even likely, that Diaz will rebound in a major way become a consistently dominant top 10 reliever for years to come.

Truthfully, though, it’s entirely possible that he won’t. How many of us would truly be surprised if Diaz’s poor performance persists, or if he turns out to have a somewhat serious injury? That sounds harsh, but volatile performance and injuries seem to be the way of relief pitchers sometimes. The data presented and analyzed today seem to back up that suspicion, at least to some degree.