Tuesday, April 9, 2019

Predicting Today's Future Hall of Fame Closers

I recently ran my annual look at which hitters and starting pitchers are on pace for the Hall of Fame, and in them, I mentioned the idea of switching things up and looking at relief pitchers as well. After all, there’s been something of an explosion in relief talent in Cooperstown over the last two years; after being stuck at five closers in the Hall for a decade, we’ve seen three new ones added in Mariano Rivera, Lee Smith, and Trevor Hoffman. Plus, Billy Wagner continues to hang around the ballot, and may see his support continue to climb in the coming years.

Now obviously, eight closers still isn’t anywhere near the body of evidence we have to work with when trying to predict position players or starters, but it might be fun to play around with, and see if we can spot any sort of nascent trends in who the Hall is electing.

It’s not a lot to work with, but there are some fun things I noticed while looking into things:



It’s really hard to “track” the Hall of Fame path for relievers
Or at least, it is in the way or to the extent that I do it for hitters or starters. Normally, my methodology for this is to look at the median Wins Above Replacement for the larger group at each age of their careers, starting in their young 20s. Part of this is what led me to realize how important it is for starting pitchers to stay good into their 30s.

It’s probably just a fluke, due to the small number of electees we have so far, but it seems that this finding is even truer for closers. Sure, Rollie Fingers, Lee Smith, Bruce Sutter, and Rich Gossage all debuted at young ages and moved permanently into their bullpen roles pretty quickly, but they’re only half of the story.

Dennis Eckersley derived a lot of his value from his closer days, but he spent the first twelve years of his career as a starter. He didn’t become a closer until his age 32 season, where he would stay for another twelve years. Mariano Rivera and Trevor Hoffman didn’t have the late-career role shuffling that Eckersley did, but both debuted relatively late by Hall standards, both starting in the Majors at 25 and taking a year or two to grow into their roles. Also like Eckersley, both stayed good into their early 30s.

They ain’t got nothing on Hoyt Wilhelm, though. If you don’t know or had forgotten, Wilhelm was a knuckleballer who got something of a late start between the skepticism being a knuckleballer engenders, as well as a delay to his minor league career to serve in Europe for several years during World War II. As such, Hoyt finally debuted in the majors in 1952, just three months shy of his 30th birthday. He would go on to pitch for over two decades, hanging things up just two weeks shy of his 50th birthday.

So yeah, tracking relievers from age 20 on would be kind of hard when half of our reference points wouldn’t even be in the bullpen for another half-decade or more, or when half of them continued racking up numbers into their 40s (and 12.5% of our sample nearly made it to 50). Like, yeah, pitching well into your later days is a big deal, but that seems just a tad extreme.

As such, we’re going to have to reduce our scope a little bit.



WAR may not be the best stat to measure relief pitchers
Or, it might be. This is still something of a contentious area among people, and there are a lot of competing factors in play

You see, a big part of Wins Above Replacement is playing time, which for pitchers, means innings pitched. And on top of that, WAR is totally context-neutral, meaning a 3-strikeout top of the first counts the same as one in the bottom in the ninth with a run-one lead. Both of those run directly counter to how closers are used in modern baseball, since even the most-used relievers wind up with under half the innings that starters get under the premise that their role in higher leverage situations requires greater intensity.

Is that fair? I’m not totally sure. I think it mostly is, although I sometimes go back and forth, and occasionally I’ll read a new study that pushes me more in one direction than the other. But regardless of whether it’s an accurate description, it definitely does line up with how Hall of Fame voters have viewed the position.

That’s not to say there’s no overlap. For instance, Eckersley leads all relievers in WAR at 62.2 WAR, although a lot of that (nearly two-thirds) comes from his days as a starter, and with over 3000 innings, he has a lot more playing time than most other relief pitchers. And Mariano Rivera is second at 56.3 (there are no catches there, Rivera is just really good).

But after those two, we have Wilbur Wood as the only other reliever with over 50 WAR (52.1). Wood was from a different era and shifted between roles; he started nearly 300 games with just over 300 more in relief, and with almost 2600 innings, only Eckersley has more innings pitched while coming out of the bullpen in over half of his games played.

The rest of the top ten by WAR is a mixed bag. Wilhelm and Gossage got to over 40 WAR threw large inning counts. Then there’s Tom Gordon just over 30 WAR while starting nearly 200 games, then Bobby Shantz (164 starts) and Greg Swindell (260) right behind him. John Hiller are virtually tied with those two, but Hiller was a full-on modern closer with just over 1000 innings. Meanwhile, Lee Smith is twelfth in WAR, Trevor Hoffman, Billy Wagner, and Rollie Fingers fall fifteenth through seventeenth, and Bruce Sutter is way down at 32nd.

That’s not a bad correlation with Hall voting, but what if there was a value stat that did look at leverage? Wins Probability Added does that, assigning win probabilities to each game state and crediting players depending on whether their plays increase or lower that probability. And relievers actually can hold their own against starters in WPA, if they’re good enough; Rivera is fourth, Hoffman is 25th, and Gossage, Joe Nathan, and Dennis Eckersley, and Billy Wagner all fall in the top 40 among pitchers.

And as you’ve probably picked up, there’s a slightly better correlation with Hall of Fame voting, with the elected closers falling coming in first, second, third, fifth, seventh, fifteenth, 24th, and 38th. Wagner, who has gotten more support than pretty much any closer who has yet to be elected, is sixth. I’m curious to see how Joe Nathan, who’s fourth, does when he hits the ballot in 2022; he had a run that was among the all-time bests for a closer, and managed a solid save total (377), but I don’t know that he tops Wagner in anything. If Billy is about to go in by then (or has somehow already made it), then Nathan will probably be the best unelected reliever, but if his case has stalled out, it will be hard to justify Nathan getting more support.

There are also something like tiers for WPA, as well. Rivera laps everyone at 57.474, then Hoffman, Gossage and Nathan are all just over 30, then Eck, Wagner, and Wilhelm are all over 27 WPA. There’s another bit of a drop after than, down to Francisco Rodriguez just a bit over 23. Lee Smith is at 20.620, Rollie Fingers is at 18.189, and Bruce Sutter is at 14.743. So if we were to try and demark milestones for Hall of Fame consideration, something in the 27-30 range seems like a good starting point.



Which active pitchers are building towards ~27-30 WPA?
The active leader in Win Probability Added among relief pitchers is Craig Kimbrel with 21.889, which is over three and a quarter more than the next closest reliever and only behind seven starting pitchers.* It also already places him thirteenth among all relief pitchers, just between Tom Henke and Keith Foulke.

*Those seven being Clayton Kershaw (40.418), Zack Greinke (29.245), Justin Verlander (28.265), Max Scherzer (25.373), Felix Hernandez (25.311), Cole Hamels (22.679), and CC Sabathia (22.308). That group definitely seems like it will be heavy on Hall of Fame inductees going forward.

Not only does that hold up well against modern starters, it also stacks up pretty well historically among relievers. Through their age 30 season (Kimbrel’s age last year), only three relief pitchers have topped 20 WPA: Gossage at 27.787, Kimbrel, and Francisco Rodriguez at 20.706. Unlike those two, Kimbrel debuted at 24, meaning that Kimbrel managed that total in a lot less time than either of them (K-Rod had nearly 200 more games and innings, while Gossage only had 30 more games but over 600 more innings between attempts to make him a starter and heavy usage on relievers in his era). It even looks good against starters, with Kimbrel ranking 24th all-time among all pitchers through age 30.

I’ll also add that it’s really weird that Kimbrel has been unable to find work, considering how good he’s been and how relatively young he still is, especially in an era that many claim is starting to be defined by strong bullpens, and in which several of this year’s supposed-contenders are clearly struggling in the late innings. He’d be an improvement to basically any team (even ones with strong relief corps like the Astros), and hopefully this unintentional hiatus ends soon and Kimbrel can pick things up right where he left off in the postseason. Just like with Wins Above Replacement, WPA can drop with bad seasons, so even if Kimbrel continues to move up the ranking, it doesn’t guarantee anything. Plus, I don’t really have a good idea of the “standard” aging curve for relievers, so I don’t know what to expect the rest of the way, but it seems like he could very easily finish his career fairly high up the leaderboard.

Moving on: I honestly forgot that Kenley Jansen and Aroldis Chapman were the same age as Craig Kimbrel; it feels like they all sort of staggered into the league at different times, with Chapman being a big name in baseball circles even before he signed and Jansen taking a few years to become a household name. Nevertheless, all three of them debuted in 2010, they’re all between 421 and 467 games played (excluding playoffs), and they’re first, second, and fourth among active relievers in Win Probability Added.

Jansen is a little over 3.25 wins behind Kimbrel at 18.531, while Chapman is just under 3 wins behind Kenley at 15.898. On the one hand, I’m tempted to say that gap isn’t so big, but on the other hand, the line between “strong Hall candidate” and “indifferent shrug from the electorate” seems to be about 4 wins, so that 3-win gap is actually a pretty huge deal. In any case, while neither Aroldis or Kenley feels like they’re in as strong a place as Craig, Cooperstown definitely isn’t out of the question for either. They’ll just need to keep it up for a couple more years to strengthen their case; to compare it to starting pitchers, if Craig Kimbrel is the Clayton Kershaw or Chris Sale of relievers, these two feel more like the Madison Bumgarner or Jacob deGrom tier.

I skipped over David Robertson, who slots in between Jansen and Chapman at third place with 17.393, but I feel like he’s a step below them. A big part of that is age; he’s three years older, so there’s less time for him to pull that total up. But some of it is also quality; Robertson has well over 100 more games than any of those three, yet still has less WPA than two of them, has fewer top-ten finishes among WPA leaders than most of them, and generally has way fewer of other recognitions (for instance, Robertson only has 1 All-Star selection to Kimbrel’s 7, Chapman’s 5, and Jansen’s 3). On top of that, he’s not going to finish with a big saves total thanks to his years as a set-up man for Rivera, Chapman, and others. It wouldn’t be unheard of for Robertson to keep it up (that is, after all, generally what separates the Hall of Famers from the also-rans), but it’s also far from certain, and his start this year definitely is not inspiring.

There are a decent number of older relievers after that with WPA totals that are solid, but still pretty far from Hall consideration. Joakim Soria is in his age 35 season and has 15.794 WPA, while Tony Watson and Mark Melancon are 34 and sit at 14.582 an 14.461, respectively. I’d put their chances all at pretty close to zero, but feel free to adjust those upwards if you think they have a late-30s run in them.

The only remaining relievers on this list with a decent total of WPA and age somewhat on their side are, in my mind, Zack Britton and Cody Allen. Britton is an interesting case, for several reasons. For one, he’s the same age as Kimbrel, Jansen, and Chapman, entering his age-31 season. Despite that, he actually has way more innings, in part because he’s a more traditional sort of closer (that is to say: a failed starter). From 2011 to 2013, Baltimore tried Britton in the rotation, and it just didn’t hold, with him running a 4.77 ERA over 254.2 innings and 46 starts.

Also, as you may have guessed, those mediocre numbers came with a net negative WPA, at -2.1. That means that his short time as closer has actually been more impressive than most, with 14.9 WPA coming in just under 300 innings (counting the start of this year). That’s actually more on Kimbrel and Jansen’s level, and like them, Britton also has a season leading the league in WPA, two other top-ten finishes, and a Reliever of the Year Award (all of those are more than Chapman or Robertson can say). The later start still hurts a bit, but I think overall, I’d actually put him on level with Chapman; he still needs a good finish to seal the deal when it comes to Cooperstown, but it only needs to be Billy Wagner/Dennis Eckersley/Trevor Hoffman good, rather than Mariano Rivera/Hoyt Wilhelm levels of amazing. He’s sort of a pleasant surprise and change of pace, after all the ones I’ve covered so far being star relievers from the get-go.

In comparison to that, Cody Allen’s case is much, much more simple. He’s in his age 30 season as has 11.257 WPA and 150 saves after basically just 6 full seasons; double that and you have a 35 year old with 300 saves and nearly 23 WPA. Granted, Allen hasn’t really demonstrated the peak that some of the others have, and whether he can duplicate his first six years as a closer is an open question given that even last year seemed like a step back, but it definitely feels like more to go on right now that Melancon or Soria, so I might as well bring it up.



On the whole, four potential Hall of Famers feels about right given that there are about 30 closers, and that’s going to make up the large bulk of relief pitchers who get votes. And all four definitely won’t make it, but zero making it also feels unlikely given what we know; attrition to about 2 staying on pace seems intuitive to me, if nothing else. We’re probably still several years away from this developing into a hard-and-fast standard like we have for hitters and starters, but it should be fun to keep an eye on at least.

No comments:

Post a Comment