Seung Hwan Oh and the fleeting nature of great relief performances

Seung Hwan Oh had a fantastic 2016 season. Relative to other age 33 seasons, it was even historic. Here’s a top ten by WAR of age 33 reliever seasons since 2000:

It’s pretty much always a good thing for a reliever to share the top of a leaderboard with Mariano Rivera. It’s also great to be reminded of Pat Neshek’s completely unforeseen and awesome 2014. Oh’s tremendous season came just at the right time, with Trevor Rosenthal having plenty of issues in the first half of the year.

With Oh holding down the ninth, the addition of Brett Cecil, a possible bounce-back from Trevor Rosenthal, as well as the continued presence of Matt Bowman and Kevin Siegrist, the Cards’ bullpen looks pretty solid going into 2016. In fact, it projects as the fifth best bullpen in baseball at the moment.

After watching such a splendid season from Oh, it’s easy to count on him again in 2017. His dominance may have even allowed you to forget that he’s still only been in the majors for a year. Of course, he racked up good numbers in Japan and Korea beforehand, but it’s still not easy to translate those numbers to Major League results. Oh is in a position where he’s not some journeyman that just came out of nowhere like Neshek, but he also just hasn’t had the opportunity to build the track record of some of the more dominating bullpen arms in the league.

So I decided to find similar reliever seasons to Seung Hwan Oh’s 2016. Because of the aforementioned dominance among 33 year olds, I decided to not use age as factor. What I did was this: I created a simple method which I called a “Similarity score”. For each reliever season over 30 innings since 2006, I took the absolute difference between that season and Oh’s 2016 for three stats: K%, BB%, and HR/FB%. I then added those three up to get the pitcher’s Similarity score, with the lowest score being the most similar. Using this method, here’s the Top 20 most similar seasons to Oh’s 2016:

These seasons as a whole are good company for Oh’s 2016. Jake McGee is the only player with three seasons on this list. Huston Street made it two times. Trevor Rosenthal is the only other Cardinal representative.

While these are some great seasons, they unfortunately didn’t always have great follow ups to those seasons. For starters, three of the matches didn’t even pitch over 30 innings in the following season, and thus won’t be shown below. Of the remaining 17, here’s what they did the following seasons:

The average of these 17 players in their first seasons are very close to Oh in K%, BB%, and HR/FB%. In the following season, they lost on average 6 points of K% and 2 points of BB%. On average, these 17 gained a full run on their FIP! Of course, when we’re talking about fantastic seasons, generally we’re going to see worse performance in the next. That’s just regression. When a player has a career year, they typically have a worse one the following season. If we’re just looking at FIP than we’re of course setting ourselves up for a lot of regression, as these pitchers all did better in terms of FIP than xFIP, which is more predictive. Still, on average this group added 67 points from their first year xFIP to the following season’s FIP.

The magnitude of this result is surprising. The projections are not quite so hasty. Steamer is pretty close to this at 3.11 FIP, but ZiPs is much more optimistic, forecasting a 2.58. I’m not going to say we should throw out the projections and replace it with the results of this exercise. Both projection systems are built on a multitude of factors, rather than a simple method of following up on 20 similar seasons. But we don’t get to see the guts of ZiPs and Steamer, and it might help understand the magnitude of regression possible if we can see 20 sequels to similarly dominating performances.

The chief takeaway of course is that with some notable exceptions, great relief pitching is fleeting. It’s inconsistent. While Oh dominated in 2016, there’s a decent chance he’s simply a good reliever in 2017, rather than one of the best in the league. If he repeats his 2016 performance, he’ll again be at the top of leaderboard with Mariano Rivera, this time for best reliever seasons from a 34 year old. That’s far from a sure thing though. Just one pitcher featured here improved their performance the following season, and only two others were able to lose less than half a run on their FIP. By contrast, seven players added more than a run and a half to their FIP.

Another caveat is this is only analyzing results, not underlying skill level. A more in-depth Pitch f/x analysis of Oh’s repertoire compared to the relievers above - one that you’re used to seeing from Joe - may lead one to be more optimistic than the results here suggest.

This does however, shine light on the idea of trading Seung Hwan Oh with one year remaining on his deal. The idea would be to trade him while his stock is likely at its highest. The Nationals are still rumored to be looking for a closer, though at this point it’s looking more and more likely they’ll go into the season relying on their internal options. Such a discussion is only purely hypothetical though: There’s little precedence for the Cardinals to trade such an important win-now piece for prospects while competing. The closest you can get is when the Cardinals made the David Freese trade, but that was trading from depth to make room for Kolten Wong, and the primary piece going back to the Cards debuted in the majors just a season later. It’s still an intriguing option though, as the actions in the reliever market indicate it would bring in a nice prospect haul.

No, I’m not going to be shouting “Doom!” from the rooftops when assessing Oh’s upcoming season, or calling the Cardinals Front Office dumb for not trading Oh this winter. Still, if you were expecting a similar season to last year from Oh, you’re probably going to be disappointed. Most here understand the concept of Regression to the Mean, but it might help to see an example of it in action. This particular example was not very kind to Cardinal fans’ expectations. Now let’s all hope Oh can prove this simple model completely wrong in near future.