Just below are a pair of 2019 statlines for a couple of high-profile relief pitchers. Think on it for a minute, and given this info, try to decide which arm you’d rather have in 2020:
- Player A (58 IP): 3.07 xFIP, 2.63 SIERA, 2.95 DRA, 39 K%, 8.7 BB%
- Player B (57 IP): 2.86 xFIP, 3.10 SIERA, 2.57 DRA, 36.2 K%, 10.6 BB%
Pretty much a toss up, right? Or at the very least, a case can be made for either pitcher. See, the stats above (xFIP, SIERA, DRA) are all predictive stats, or metrics that are designed not to evaluate the results of a pitcher’s performance, but rather to determine the results a pitcher should achieve given his isolated performance. The subjects of the above comparison include the 2018 AL Reliever of the Year, current Mets closer Edwin Diaz (Player A) and the reigning AL Reliever of the Year, Yankees lefty Aroldis Chapman (Player B). These two relievers had seasons that, based on their peripheral stats, should have netted them similar results, right? Well, let’s take a look:
- Diaz: 5.59 ERA, 4.51 FIP, 1.38 WHIP
- Chapman: 2.21 ERA, 2.28 FIP, 1.11 WHIP
Not quite as similar this time, as the Yanks’ closer posted results that any reliever would kill for, while a couple boroughs over, Diaz was being thrashed by the New York media night in and night out. And, to be fair, he was an easy target — in the offseason the Mets had dealt top prospect Jarred Kelenic to acquire the elite closer, only for him to post results that more closely mirrored that of below-average Pirates righty Clay Holmes (5.58 ERA, 4.97 FIP) rather than a top tier bullpen arm.
This data is all very interesting, but specifically, leads me to ask two questions:
- Why is there such a discrepancy between Diaz’s expected and actual statistic outputs?
- Should we be using predictive or result based analysis when evaluating the quality of a player’s season?
Casual observers, seeing the near 4 run uptick in Edwin Diaz’s ERA from 2018 to 2019, would assume that something drastic had changed from one season to the next. So, let’s find out.
The only moderately significant difference in Diaz’s stuff from 2018 to 2019 would be a slight dip in his fastball’s vertical movement, but aside from that, his fastball/slider combo remained elite, actually seeing slight increases in both velocity and spin rate.
Some more noticeable changes were the sharp increases in Diaz’s Hard Hit %, xwOBAcon, and Barrel %. So, is it as simple as that? Diaz’s stuff hadn’t worsened, but hitters just caught up to him? Not quite. And while it would be nice if it were that simple, the correlation between hard contact and run prevention isn’t always as strong as one would think. Not only because the aforementioned batted ball metrics can be fluky with inordinately high variability (especially in sample sizes as small as a single season), but also because with a K-BB rate as high as Diaz’s (38.2% in ‘18, 30.3% in ‘19), it’s easy to get away with allowing some hard contact. With similarly poor Hard Hit %, top tier arms like Hader, Nick Anderson, and Shane Bieber all excelled at preventing runs in 2019. Unsurprisingly, these three pitchers were also masters of the strikeout, recording K rates of 47.8%, 41.7%, and 30.2%, respectively.
So, if an increase in Hard Hit % isn’t the sole culprit of Diaz’s seeming outlier of a season, then what is? Well, it’s exactly what has plagued pitchers for decades: defense and luck.
In 2018, Diaz’s Mariners posted a collective DRS total of -23. Not great, certainly nothing to be proud of, but you could do a great deal worse. Conveniently, after being dealt to the Mets prior to the 2019 season, that’s exactly what they did, posting a horrendous -93 DRS, trailing only to the Orioles’ -105 in terms of putridity. Even Baseball Savant’s Infield Outs Above Average, which places the two teams a bit closer, has the 2018 Mariners at -11 OAA, while the Mets posted -13. So yes, defense played a crucial role as well, but still, it’s more than that.
In addition to an utterly insane .377 BABIP, Diaz’s 26.8% HR/FB in 2019 suggests a supremely unlucky season, as pitchers have an alarmingly low level of control over whether fly balls actually fly out of the yard. This is why xFIP, which regresses a pitcher’s HR/FB to league average (as opposed to standard FIP), shows Diaz in such a positive light.
And no, I don’t mean to just re-explain what has already been discovered by Voros McCracken and further explored by dozens more, but it is important to contextualize seasons such as Diaz’s, because if we don’t, it’s easy to get swept up in warped narratives, be it positively or negatively.
Now, I hope I’ve proven that Diaz suffered a far worse fortune than he deserved—for a variety of reasons—and that he’s still an elite talent, but the question still remains, did Diaz have a good season in 2019? Personally, I would say that he did. Edwin Diaz consistently pitched well and gave his team a very real shot to win, but due to a lack of luck and defensive talent, he was rewarded with poor surface level results.
I’ve seen some debate on Twitter recently about whether it is fair to evaluate a player’s season based on predictive statistics. People I greatly respect have been on both sides of this debate, with some saying that regardless of how a player “should have” performed, the results speak for themselves. That it is unreasonable to exclude unfiltered result-based analysis in favor of peripheral statistics. Conversely, supporters of predictive stats would retort by saying that further isolating how a player is truly performing is far more indicative of the quality of his season, regardless of how luck or teammate performance factored into the results. Personally, I’d side with the latter, though I understand the position of the former. Diaz may not have seen impressive results on the surface, but he pitched nearly as well as any reliever in the league, and was let down by a series of factors not within his control. It’s unfair to punish him with disingenuous talk of a “down year” when he was, by all statistically revealing measures, one of the best arms in the game.
It may seem like I’m getting caught up on semantics here, but even in a game so intensely dominated by numbers, language and discourse is important, so it’s crucial to know that we are operating on the same plane. And, the first step we must take to get to that plane includes acknowledging that it is impossible for Edwin Diaz to have a comeback year in 2020, because there is nothing to come back from. The right hander will almost certainly show out again in 2020, and the results will reflect it.
Featured Photo: Brad Mills, USA TODAY Sports