PECOTA Hates Corey Kluber, And That’s OK

Or: How I Learned To Stop Worrying, And Take Corey Kluber Projections For What They’re Worth

Corey Kluber is the Cleveland Indians ace, a reigning Cy Young winner and was named Cleveland’s Professional Athlete of the Year for 2014. He’s cheap, on the right side of age 30, and under team control for a few more years. What’s not to love?

More from Cleveland Guardians News

Well, January is (traditionally) when MLB’s hot stove season dies down, and it becomes “Best Shape of His Life” (BSOL is often the Twitter shorthand) season, when guys are in “the best shape of their lives” and ready to come in and dominate this season. Hope runs abound, predictions are rampant, and a few of the major baseball blogs and sites put out their annual projections on every single player in baseball. To name a few, there’s Steamer, whose projections can be found on Fangraphs, there’s ZiPS, developed by the Baseball Think Factory, Brian Cartwright’s Oliver at The Hardball Times and PECOTA, a pretty complex projection system developed by a pre-FiveThirtyEight Nate Silver at Baseball Prospectus.

On the surface and to many fans, these projections are nothing more than a nerdier and more objective version of ESPN’s NBARank popularity contest: there’s no awards for being the best projected player, and they can be very wrong at times. But to the more intense

(or nerdier, if you want to be a jerk about it)

fans, these projections are a fascinating look at how the numbers expect any given player to perform.

I included the links to how these projection systems work (at least on a basic level), but let’s focus on PECOTA, which explained the system’s bare bones in the BP glossary:

"Stands for Player Empirical Comparison and Optimization Test Algorithm. PECOTA is BP’s proprietary system that projects player performance based on comparison with historical player-seasons. There are three elements to PECOTA:1) Major-league equivalencies, to allow us to use minor-league stats to project how a player will perform in the majors;2) Baseline forecasts, which use weighted averages and regression to the mean to produce an estimate of a player’s true talent level;3) A career-path adjustment, which incorporates information about how comparable players’ stats changed over time."

PECOTA is an algorithm, and therefore can’t actually “hate” anything; it can, however, have no idea what to do with a player. BP will release PECOTA’s full projections on Monday, January 26, but they’ve been teasing morsels of info the past few days, and one such nugget fell into Indians fans laps yesterday (emphasis mine):

"The Cy Young race between Corey Kluber and Felix Hernandez was one of the more compelling award finishes in recent memory. Two of the American League’s finest hurlers pitched at the top of their games all summer long, dominating opponents on the scoreboard while excelling in the three true outcomes categories. Voters ultimately favored Kluber by the narrowest of margins, but the battle spawned the best of baseball’s award season: a competitive race, thoughtful reflections, and well-reasoned discourse. As soon as I got access to the new PECOTA’s, I quickly glanced through the pitchers, figuring that these two would again be neck and neck near the top of the system’s leaderboards.Not exactly. Hernandez is right where we all expect, projected to finish third in WARP among all pitchers, well behind Clayton Kershaw but considerably ahead of nearly everybody else. Kluber, however, isn’t anywhere in the top 10. Or top 20. Or top 50. Or top 200. To find Kluber’s name, you have to scroll past Daniel Hudson, below Ben Lively, and even beyond someone named Preston Guilmet. All told, there are 159 names above Kluber who, like potential rotation mates Gavin Floyd and Zach McAllister, is projected to accrue less than a full win in 2015. Barring arm surgery in early May, I feel pretty comfortable taking the over on that one."

Oh, former Indians reliever Preston Guilmet? That’s … not encouraging. (Hilarious side note: Guilmet was claimed by the Toronto Blue Jays, along with former Cleveland roster fodder Scott Barnes.)

Aug 8, 2013; Cleveland, OH, USA; Cleveland Indians relief pitcher Preston Guilmet (36) pitches during the third inning against the Detroit Tigers at Progressive Field. Mandatory Credit: Ken Blaze-USA TODAY Sports

For what it’s worth, Steamer projects Kluber at a lower-than-last-season-but-still-excellent 4.0 fWAR, with Kluber making 30 starts (192 IP) and a solid 3.21 ERA (and an unsurprisingly better Fielding Independent Pitching of 3.05). ZiPS is even more bullish, projecting 31 starts (approximately 206 IP), and a 2.97 ERA (3.03 FIP).

So, while we don’t know PECOTA’s exact projections yet, it’s fair to say there are other projections that DO think Kluber will be excellent in 2015. PECOTA is probably putting a heavy emphasis on his poor minor-league numbers, and likely sees Kluber’s past two seasons as more anomalies than long-term development.

And BP realizes much of this as well. Brendan Gawlowski, author of the blurb for BP, quickly puts out PECOTA’s fire in his next paragraphs (emphasis again mine):

"Kluber, of course, presents a challenge for all forecasting systems. Projections have a difficult time evaluating breakouts, particularly for someone like Kluber who arguably broke out twice. PECOTA expects significant regression in strikeout rate and workload, which makes plenty of sense, but in groundball percentage and home run rate as well, which is a little tougher to see. Kluber also pitches in front of Cleveland’s porous defense, which has a cascade effect on the rest of his line, affecting expected hits allowed, BABIP, WHIP, innings, and so on.Strangely though, PECOTA doesn’t think it’s out of the question for Kluber to improve upon last season’s performance. If we lump him in with the top 30 pitchers in projected WARP, the system thinks he’s one of the more likely players among that group to break out and the least prone to collapse. And remember, that’s based off recent performance, not the system’s pessimistic 2015 projection.What looks like hedging is really just PECOTA throwing its algorithmic hands up. Twenty-six-year-old rookies with a history of control problems don’t suddenly turn into Cy Young winners very often, and the system is having a hard time reconciling Kluber’s sensational 2014 season with the rest of his profile. Whether you want to see this projection as a triumph of the human element or a warning that pitchers who find magic at 28 may lose it at 29 is up to you. Regardless, this is undoubtedly one of the most controversial projections of the season, one that adds another layer of intrigue to Kluber’s defense of the AL Cy Young award."

Sep 22, 2013; Cleveland, OH, USA; Cleveland Indians starting pitcher Corey Kluber during a game against the Houston Astros at Progressive Field. Cleveland won 9-2. Mandatory Credit: David Richard-USA TODAY Sports

It’s worth nothing that neither PECOTA, ZiPS nor Steamer tries to factor in any sort of health risk to their projections; if a player has played/pitched a certain amount in the past, the systems often have to assume the same going forward. The inverse is often true: if a guy has a small sample, the systems generally assume they’re not going to suddenly get more playing time (or, in the case of ZiPS, can’t factor in for a guy such as Carlos Carrasco, who spent much of last season and parts of his career in the bullpen, but figures to be a starting pitcher from Opening Day for 2015).

Both of these can be slightly scary caveats, particularly in the case of Kluber. Gawlowski nails the Kuber Conundrum pretty succinctly; this is a guy who (to his immense credit) has put in the work and raised his profile from “Maybe he’s a back-end starter” (2011) to “this is a solid-to-excellent mid-rotation guy” (2013) to the brilliance of his 2014 campaign. Don’t believe anyone who says retrospectively they saw Cy Young winner when the Indians traded for Kluber in 2010 (and, if there were people who saw an ace in Kluber, Cleveland was probably not getting him for a half-season of Jake Westbrook). Projection systems often struggle to define adjustments (such as adding a two-seamer in 2011) like Kluber has made in the past three years. And no projection system can accommodate for the hard work Kluber has put into his delivery (and others have noticed, Doug Thorburn of BPro said this offseason Kluber had the best delivery mechanics in the AL Central). Both the two-seamer and the delivery suggest real, tangible changes (i.e., NOT BSOL) which a projection system wouldn’t (and right now, couldn’t) know.

Thinking about PECOTA and other projection systems, I’m oddly reminded of late film director Stanley Kubrick, one of my favourite directors but also admittedly one of the more challenging directors to watch. Kubrick’s films, while enjoyable, almost always demand a crazy amount of patience from the viewer; he takes his time with shots, shows an insane, almost oppressive amount of detail, and isn’t here to entertain you. Kubrick (usually) has something to say, and if you’re not listening and paying attention, you’re wasting your time.

While my favourite Kubrick films remain “2001: A Space Odyssey” and “A Clockwork Orange” (both loose adaptations of novels), reading the MLB projections strangely reminds me of another classic Kubrick work (also very loosely adapted from a novel), his 1964 political satire “Dr. Strangelove or: How I Learned to Stop Worrying and Love the Bomb,” which features some of the late, great Peter Sellers’s finest acting.

Aug 15, 2014; Cleveland, OH, USA; Cleveland Indians catcher Yan Gomes (10) congratulates Cleveland Indians starting pitcher Corey Kluber (28) after Kluber struck out the side during the fifth inning against the Baltimore Orioles at Progressive Field. Mandatory Credit: Ken Blaze-USA TODAY Sports

If you’ve never seen it, it’s a black and white “dark” comedy (“nightmare comedy” in Kubrick’s words) with Sellers in multiple roles, and takes aim at the Cold War and the terror of nuclear weapons during that era. Kubrick intended to write the film (an adaption of the novel Red Alert) as a drama, but when he actually began work on the screenplay, he began to see the humor in the grim nature of nuclear weaponry and the absurdity of many aspects of the Cold War and felt he had to write the film as a satire. In many ways, “Dr. Strangelove”is Kubrick’s “eff you” to the terror of a nuclear winter, and it’s one of the reasons the film still has relevance today, even with new threats to our safety.

So, uh, what the hell does that have to do with projections for Corey Kluber?

These projections all have their faults (and to the credit of the creators, they all are usually quite transparent regarding what their systems do well and what they do poorly), but they’re certainly useful. If nothing else, looking at as many of these projections and being aware of both their strengths and faults can be an invaluable challenge to our expectations of a player going forward. Information isn’t inherently good nor bad, it’s more in how the information is used where morality comes into play; in the case of Corey Kluber’s projections, we have ZiPS and Steamer (among others) to get us jazzed, and then PECOTA (likely) comes in to smack us in the face with baseball reality: “Yeah, he could be great, but it could not be real.” And if you aren’t paying attention to the what the projections are telling you (and HOW they’re telling you), you’re wasting your time. PECOTA is telling us, effectively, they don’t have a whole lot to work with on a rare breed such as Kluber.

Reading them can be grim, but it’s also strangely comical. I love learning about advanced metrics in baseball, it’s helped my understanding and enjoyment of the game immensely, even if I’m still terrible at math. But one area of growth that’s been key for me is understanding that these numbers aren’t set in stone, and certainly need context, and I sort of find the entire exercise “funny” as a result.

Nov 19, 2014; Cleveland, OH, USA; Cleveland Cavaliers mascot Moondog introduces 2014 American League Cy Young award winner Corey Kluber of the Cleveland Indians during a game between the Cleveland Cavaliers and the San Antonio Spurs at Quicken Loans Arena. Mandatory Credit: David Richard-USA TODAY Sports

This isn’t my “Dr. Strangelove;” there’s no “eff you” to the projection systems. Please, keep publishing them, keep working on them, keep making new and better ones. Baseball analytic methods continue to grow and improve, and perhaps one day, they will be able to factor in injury risks, such as perhaps a pitcher throwing nearly 100 more innings than he ever has in his career one year (though important to note, I’m NOT talking about the Verducci/Year After Effect, which doesn’t apply to Kluber as he’s older than 25 and it’s a really poor method, but the Indians are monitoring him regardless), or a pitcher being able to make adjustments and develop. MLB players usually don’t arrive to The Show as a finished product.

But until they can, we as fans (and especially those of us who occasionally write publicly about baseball) need to challenge ourselves with the projections, and not go to one extreme of thinking Kluber’s 2014 is a unrepeatable flash in the pan to the other extreme of guzzling the Kluber Kool-Aid (which, admittedly, we’ve been pretty drunk on at Wahoo’s on First for a while). We don’t know the truth yet, but here are some estimated guesses. It could be sweet, it could be crap. But like Peter Sellers in “Dr. Strangelove,” it’s probably best to be laughing hysterically regardless.

PECOTA Hates Corey Kluber, And That’s OK

Or: How I Learned To Stop Worrying, And Take Corey Kluber Projections For What They’re Worth

More from Away Back Gone

Schedule