August 10, 2015
What Are the Different Pitcher WARs Good For?
This piece is more exercise than thesis, so I won't waste much time introducing it. Below, you can see the top 20 pitchers of 2015, as estimated by each of the three major statistics we use to peg pitcher values. I've highlighted the players who appear on just one of the three lists, because while it's notable that this gives us a head start toward establishing consensus tiers of the league's top pitchers, the meat of this exercise is to search for pitchers of whose quality we're less certain. (Please note that all of these stats are as of last Thursday. The changes since then are small, and the cutoff had to happen sometime. As you'll see, the idea is to talk about what factors lead certain pitchers to be evaluated differently by different models more than to make concrete statements about anyone in particular.)
The fWAR Darlings
Carlos Carrasco: 12th in fWAR (3.4), 22nd in DRA PWARP (2.7), 32nd in bWAR (2.3)
There's no great mystery about Carrasco: He's a good pitcher, and the things he does best (striking out a lot of people and issuing very few walks) are the things fWAR values more highly than anything else. This year, he's striking out over 27 percent of all opponents—more of them than he did even during his breakout campaign last year—and walking just 5 percent of them. However, he hasn't retained the newfound abilities to keep the ball on the ground and get a lot of outs on balls in play.
Those things had never been big parts of his skill set in the past, and showed up in a season split between the rotation and the bullpen. The smart bet is that he's among the class of pitchers who will always allow above-average BABIPs, and who are therefore overrated (however slightly) by FIP-centric value metrics. With his dominant strikeout numbers, though, it's impossible to argue that he's anything less than a solid second-tier starter, as shown by his placement not far outside the top 20 by either of the other metrics.
Jose Quintana: 15th in fWAR (3.1), 73rd in DRA PWARP (1.4), 29th in bWAR (2.4)
FanGraphs is the site seemingly going out on a limb for Quintana, but it's DRA's evaluation of him that truly stands out. If you trust DRA, then the commonly expressed sentiment that Quintana is underrated is dead wrong. Here's what DRA is seeing that the other metrics aren't:
Jose Quintana Platoon Splits, 2015
After two full seasons of being better against right-handed hitters than against lefties, Quintana has earned the benefit of the doubt for now. It's important to remember that DRA is a measurement of value delivered, not a predictive statistic. (Its companion, cFIP, takes care of that, and Quintana's 92 cFIP places him among the top 30 pitchers in baseball.) Still, in terms of wins and runs added to the ledger to date this season, DRA feels that bWAR overrates Quintana, and that fWAR is crediting him with as much as two more wins than he's actually earned.
Clay Buchholz: 16th in fWAR (3.1), 27th in DRA PWARP (2.5), 30th in bWAR (2.4)
Buchholz might be headed for Tommy John surgery, but whatever you think of him, you have to hope he can avoid it and return at the level he found this season. There's not a lot of real disagreement on Buchholz, according to the metrics. If the list above had gone to 30 for each stat, we wouldn't be talking about him.
Since we are, though, here's what fWAR is seeing in Buchholz this season:
Buchholz switched from using his four-seamer (in 2013) and cutter (in 2014) to his sinker as his primary fastball this season. He also threw his changeup nearly twice as often as he had in 2014. Put it all together and he was filling up the strike zone like never before. He'd walked just 5 percent of opponents, a career-best mark, and he was right around his career high in strikeout rate, too. With Buchholz, health is always a question, but he'd seemed to find a consistent, comfortable repertoire this season (arguably, for the first time in his career), and was making health the only question. One can only hope that's still true when he next takes the mound.
Michael Pineda: 17th in fWAR (3.0), 55th in DRA PWARP (1.6), 74th in bWAR (1.6)
Here's a true outlier. Only fWAR thinks Pineda is a top-tier pitcher, and it's not terribly close. Pineda has a sensational strikeout-to-walk ratio this season (how does 117:15 strike you?), but has given up 129 hits in 118 innings. Thirteen of those hits have cleared the fence. It's possible that fWAR is noting Yankee Stadium's sky-high park factor for home runs and excusing those homers, and the strong overall power numbers opponents have posted. The other systems aren't as forgiving, and in the case of DRA, at least, I can point to at least one good reason for that:
Michael Pineda Home/Road Splits, 2015
Jon Lester: 18th in fWAR (3.0), 77th in DRA PWARP (1.4), 68th in bWAR (1.7)
We'll lump these two together, both because they're on this list of unique fWAR darlings for similar reasons and because the differences between them are fascinating and deserve to be set in some relief.
First of all, the thing you should know by now: These are two of the worst pitchers in the league at discouraging runners from attempting steals. Lester's inability to throw to first base is well documented. After correcting for other factors, though, Lester being on the mound only increases the willingness of a runner to try a steal by 14 percent. I say "only" because, while that's the third-highest figure in MLB this year, it pales in comparison to Ross's league-worst number: 18 percent. Now, Lester is worse at stunting opponents' success rate, with a 0.6 percent SRAA that ranks fourth-worst in the league. (In other words, when he's on the mound, a steal attempt is about six-tenths of a percent more likely to succeed than when a pitcher with average steal-prevention skills is out there.) Runners are only 0.02 percent more likely than average to be successful when they run against Ross. If neither of those numbers sound like much, though, it's because they aren't: Pitchers' effects on the running game mostly take the form of discouraging or encouraging opposing runners. That's why Ross is even worse than Lester at controlling that dimension of the game.
The difference—the reason DRA likes Ross so much more than it likes Lester, despite the weakness they share—is in the support they each get in other aspects of the game. Lester works almost exclusively with David Ross, one of the game's best defensive catchers. Ross has bailed Lester out of some tough spots, catching those runners who are always trying their luck. He's also been one of the best pitch-framers in the game. Without those things going for him, Lester would be materially less successful. DRA recognizes that, and rightly apportions the credit for those things to Ross. Now, could Lester find a way to succeed if he didn't have Ross behind the plate to aid him? One hopes so. It seems unbelievable that he wouldn't be able to, given all the things he can do on the mound. Even so, the success he's had in reality this year is inextricably tied to his teammates, and especially his catcher.
Tyson Ross, on the other hand, throws mostly to a pedestrian framer and thrower in Derek Norris. He also induces groundballs at a phenomenal rate and works in front of a shoddy overall defense. It sounds strange, and isn't really accurate, to say that DRA credits him for that, but it accounts for it. Though it must be considered in the context of his home park and weighed against his high walk rate, Ross's success has come more independently than has Lester's. Whereas fWAR does all it can to remove a pitcher's teammates from the equation and bWAR credits a hurler for finding success in step with his teammates, DRA tries to see the whole picture, but assign credit on a purely individual basis. That's how fWAR came to so much higher an opinion of both pitchers than bWAR did, and how DRA sees the two otherwise-similar hurlers so differently.
Lance Lynn: 20th in fWAR (2.8), 40th in DRA PWARP (1.7), 40th in bWAR (2.1)
Lynn has fanned over a quarter of opposing hitters this season and is walking fewer than 8 percent of them. He's also allowed just nine home runs in 121 1/3 innings. He's not a groundball guy, though, so DRA is probably observing the oddly low number of dingers he's allowed per fly ball. Both DRA and bWAR task themselves with teasing out the support Lynn gets from an excellent defense, both behind him and behind the plate (in the person of Yadier Molina). Most notable of all, though, is that a pattern is emerging: fWAR seems to value sheer volume much less than the other metrics do. Lynn joins Pineda and Buchholz as pitchers with notably lower innings counts than most of the other names on all of these lists, but who nonetheless reached the top 20 on the FanGraphs list.
The Doctors of DRA
Cole Hamels: 9th in DRA PWARP (3.9), 22nd in fWAR (2.7), 21st in bWAR (2.7)
In an ordinal sense, there's not much of a discrepancy here. Hamels couldn't have come closer to making the other top-20 lists; he's just especially beloved by DRA. I wrote about the reasons too recently to spend a great deal of time on them again here, so go read up, if you're interested. Suffice it to say that he's one of those guys for whom every contextual adjustment just seems to reveal more value.
Jeff Samardzija: 19th in DRA PWARP (2.8), 30th in fWAR (2.4), 104th in bWAR (1.2)
In his first three seasons as a big-league starter, Samardzija had a 24 percent strikeout rate and a 7 percent walk rate. This year, he's striking out only 18 percent of all opponents, but he's also walking only 5 percent of them. His 4.35 ERA isn't the result of a wacky BABIP or an elevated home run rate on flies; he's just allowing a lot of balls in play and getting below-average results on them.
Part of the reason, though, is the whopper of an opponent set he has drawn. Seven of Samardzija's 22 starts this season have come against the three best offenses in baseball: the Blue Jays, the Tigers, and the Yankees. Nine of 22 have come against teams among the 10 least strikeout-prone in MLB, including two against the contact-mad Royals. DRA elevates Samardzija to a level fWAR won't match because it adjusts for that nasty schedule.
Last but not least on the list of arguments in Samardzija's favor, and providing further evidence that DRA values volume more highly than does fWAR, Samardzija is second only to Corey Kluber in batters faced this season. That durability has value, even if attaining it has forced Samardzija to make some adjustments that hurt his strikeout rate and per-PA value. It seems as though, if the Sox have now come to their senses, there's still time for some team to claim Samardzija on waivers and offer Chicago more than a compensatory pick's worth of prospect value in order to put him at the front of their rotation.
Dellin Betances: 20th in DRA PWARP (2.7), 37th in fWAR (2.1), 23rd in bWAR (2.5)
If the season ended today, Betances's DRA- of 12 would be the second-lowest ever for anyone with 25 or more innings pitched in a season. (Baltimore's Bob Milacki had a DRA- of 1 in exactly 25 frames in 1988, which I'm sure I'll explore more some other time.) Betances has 55 innings pitched. I'm at a loss to explain what, exactly, makes him so unique. Neither his opponent set nor his aggregate park factors seem exceptional. He's striking out over 40 percent of batters, of course, which is amazing, but nothing Craig Kimbrel or Aroldis Chapman haven't bettered before. He's inducing a ton of infield popups and is suffocating to would-be power threats, but not in any way that Wade Davis hasn't bested. He's walking over 11 percent of opposing hitters. Yet DRA says he's having an utterly historic season. It eludes me, but there's no question he's been very, very good. Maybe a one-by-one breakdown of all the situations he's faced this season would show something we just can't see.
The bWAR Specials
Kyle Gibson: 15th in bWAR (3.2), 78th in fWAR (1.4), 45th in DRA PWARP (1.7)
Sometimes, it's dangerous to overthink things. Those who know these debates know that Baseball Reference's version of pitcher WAR is based on runs allowed per nine innings, adjusted for defensive support, role, and park. When a guy like Gibson, whose runs-allowed results look out of joint with his peripheral numbers, pops up as a favorite of bWAR, the safest bet is that the WAR just reflects a bit of good luck or good sequencing.
Gibson keeps the ball on the ground as well as practically anyone. He controls the running game well, which is one reason DRA treats him more kindly than fWAR. He's faced the Royals four times this season, and he works mostly with defensive nightmare Kurt Suzuki, so DRA cuts him a break there, too. Again, though, Gibson shouldn't be mistaken for a top-20 pitcher in the league under any circumstances. He is, at best, the fringe top-50 guy PWARP sees, and he's more likely in the solid-average category, where fWAR places him.
Matt Harvey: 16th in bWAR (3.1), 29th in fWAR (2.5), 25th in DRA PWARP (2.5)
Again, this doesn't really indicate a serious difference of opinion. We're working the margins here. Harvey's rates of strikeouts, walks, groundballs, and damage on fly balls have all run the wrong direction since he returned from Tommy John surgery, and the more finely tuned stats are more sensitive to those things than is bWAR, but no one is saying Harvey isn't an ace. He just has yet to recover his best form. (Maybe he never will, but it's too early to say that for certain.)
Carlos Martinez: 17th in bWAR (2.9), 50th in fWAR (1.9), 133rd in DRA PWARP (0.9)
Call this the #CardinalsDevilMagic segment. Lackey's run prevention far outruns his actual performance, thanks to a strong defense (his .290 BABIP is the second-best of his career), very few homers flying out of the park (just 1.8 percent of opponents' PAs have ended with a round-tripper, his lowest since his prime in Anaheim), and good sequencing luck (he's stranded 80 percent of all baserunners allowed). He's the boring half of this, though. Ho hum, a Cardinals pitcher benefits from that organization's combination of excellent work on mental skills, noted success at remaking veteran hurlers, and that always-sturdy set of fielders.
Carlos Martinez, on the other hand, is really intriguing. The gap between any two of the three metrics is sizable. Basically, he's been impossible to score upon, but it's not clear why, or whether it's sustainable. His strikeout rate is good, but not special, given the league context. His walk rate is bad, especially given the league context. He induces groundballs, but he also allows a relatively high number of homers. Part of DRA's distaste for Martinez is that he's struggled against left-handed hitters this season, and not because of a crazy BABIP. He's fanning fewer than twice as many lefty batters as he's walking, and those batters have a .750 OPS despite a .296 BABIP. It must run deeper than that, though, because DRA PWARP says Martinez has been a lesser pitcher than Nate Eovaldi, Jeff Locke, and Alfredo Simon. It's true that he often seems to be tiptoeing through the opposing lineup, dancing just out of trouble, living dangerously, but given his blend of stuff and name value, the conventional wisdom has been that he's breaking out and finding his true level. Maybe there's reason for pause on that front.
Taylor Jungmann: 19th in bWAR (2.8), 54th in fWAR (1.8), 34th in DRA PWARP (2.0)
This one makes less sense even than Betances. Jungmann has a 2.26 ERA in 11 starts, but the big reason that he shows up here is that it has come in what Baseball Reference estimates to be a very difficult blend of parks, and with a bad defense behind him. By bWAR's reckoning, Jungmann is preventing runs very well in a very hostile set of circumstances. Even so, the lack of a standout strikeout or walk rate and the .278 BABIP seem to point toward imminent regression.
None of this is exceptionally actionable information. It's mostly a matter of interest. The divide over the best way to evaluate pitchers with statistics is longstanding and continuing. The more time we all spend studying the places where these three points of view diverge from one another, the better we'll be able to contextualize and translate those numbers, and maybe, eventually, identify players one or another of those stats capture better than others.