The Myth of Ceilings

Tonight, the 2016 MLB draft kicks off, with the first couple of rounds being broadcast on MLB Network (and MLB.com) starting at 7 pm eastern. For the next few days, teams are going to make their best educated guesses as to what players might develop into in four or five years, and down the line, we’ll look back at these few days as a big turning point for some organization. Someone is going to hit a home run in the draft tonight, and end up with a franchise-altering class of talent that could propel them forward for years to come. Teams put a lot of resources and energy into trying to make their picks as effective as possible because the long-term impact of quality drafting can be tremendous.

But the reality is that we’re not going to know, tonight, who is hitting a home run as they’re taking their swings. The draft is an exercise in forecasting, but it’s the most difficult kind of forecasting to do; projecting long-term futures based on incomplete or unreliable information. At the very top of the draft, there’s usually enough information to provide some confidence that the players being picked are the cream of this particular crop, but beyond those first few picks, it becomes very difficult to parse the differences. This isn’t a knock on scouts or scouting; it’s just that this particular job of identifying future production so far out is just really hard.

But, interestingly, even with the tremendous uncertainty that goes along with these picks, one term has stuck in the draft and prospect lexicon that suggests that we know more than we actually do, and you’re going to hear that term a lot tonight. For almost every pick, you’re going to hear about a players “ceiling”, or upside, or some other term for the upper limit of his potential. Big guys with big fastballs are given heigh ceiling grades, and diminutive college hitters with contact skills but lacking in power will be labeled as role players or utility guys, and by and large, you’ll find a lot more high-end players in the first group than the latter one.

But when you look at the current major league leaderboard, it should become pretty clear that the idea of an actual ceiling for any player is a significant overestimation of our ability to project their skills this far out.

You know who currently leads major league hitters in WAR? Well, sure you do, because it’s Mike Trout, and by this point of every season, the answer is always Mike Trout. But do you know who is right behind him, at the #2 spot on the list of the most productive hitters this season? Jose Altuve, the Astros 5’6 second baseman who signed for $15,000 back in 2007. Never has there been a more obvious candidate for a low-ceiling player than Altuve, who is the shortest player in the big leagues, and even after he got himself to the big leagues through dominating performances in the minors, he looked like a singles-and-steals slap hitter, a guy who could play at this level because of his contact but simply lacked the power to be anything more.

But then, somehow, Altuve developed power. He’s running a .220 ISO at this point, and even if that won’t last, his power development over the last few years has put him in the running for the title of baseball’s best second baseman. And now that he’s drawing walks, he’s basically become an all-around offensive machine, and one of the very best players in baseball at any position.

Altuve, of course, is an outlier. There aren’t any other 5’6 guys in baseball because height really is an advantage, and he just happens to be the one 5’6 guy who has enough other skills to overcome his size. But there are too many other guys who got hit with low-ceiling labels who have turned into superstars for us to continue to give credence to a term that suggests that we know right where the upper limit on a player’s potential actually is.

Look at the list of the best players in baseball, and you’ll find names like Trout and Bryce Harper mixed in with Paul Goldschmidt, Josh Donaldson, Mookie Betts, Matt Carpenter, Ben Zobrist, and Kyle Seager. These guys have been MVP caliber performers despite the fact that their physical limitations had them all pegged as “low ceiling” players, and in basically every case, the reason they’ve outperformed expectations is that they figured out how to hit for power. And when you add power to a broad base of other skills, you get excellent all-around performers.

Projecting power from some players is pretty easy. Everyone knew Joey Gallo had 80 power coming out of high school, and it’s not a surprise that he’s turned into one of the most prodigious sluggers in the minor leagues. But identifying that a player not only doesn’t have power now, but cannot develop it in the future? That’s a lot harder, and something we should be significantly more reticent to suggest. Non-power hitters develop power regularly enough that it’s probably prudent to stop inferring that they can’t or won’t, and we should stop treating them as if performance levels that require power are out of their range.

This doesn’t mean that we don’t know anything about which groups of players are more likely to become high-level performers than others, nor should teams just start drafting every slap hitter they can find under the belief that they can teach them to start driving the ball. But we should probably reconsider whether the word ceiling, which suggests that there’s a practical upper limit of performance, is really the right term to describe the different probabilities of outcomes.

Carlos Correa had more star potential than Mookie Betts, but both had star potential, even if only one of them was talked about in that way. Correa’s development into a great young player was more likely, but Betts’ development into this kind of player wasn’t some kind of impossible, never-before-seen occurrence. We should be able to talk about probabilistic outcomes without resorting to such a firm idea of an upper limit. Too many players have broken through these mythological ceilings to retain that term’s credibility.





Dave is the Managing Editor of FanGraphs.

newest oldest most voted
Les Vegetables
Member
Les Vegetables

Nice piece Dave, and I completely agree. In fact I really wish all baseball forecasting was available as a probabilistic curve, and not just what the mean outcome is. Considering the amount of variation in baseball performance, there has to be a wide variety of outcomes once you factor in health/injury history, age, potential for improvement, etc. right? Does that exist anywhere? Is ZiPS/Steamer available as a probability distribution?

Klubot3000
Member
Member
Klubot3000

Yes, Baseball Prospectus does a form of this type of probabilistic projection in the form of their Breakout/Improve/Collapse/Attrition probabilities. Gives a useful sense of the probability curve for individual players. Worth taking a look at.

Matt Hunter
Member
Member

Not exactly the same as ZiPS/Steamer, but SaberSim (which uses Steamer as input) has distributions for daily projections for games and players (https://www.sabersim.com/mlb)