Kris Bryant: The Unluckiest Cub

I realize that it’s rather déclassé to refer to Kris Bryant as unlucky. The man is a millionaire, a model, a professional baseball player, and—this is, perhaps, the most infuriating part—by all accounts an extraordinarily humble and well-adjusted young man. In other words, he falls into the category of persons whom you’d least like to run into at your high school reunion, and most like to see stub their toe every once in a while. Or something like that.

But still, the fact remains: in at least one important way, Kris Bryant has been unlucky this year. In fact, he’s been more unlucky than any other qualified batter on this Chicago Cubs team. So let’s talk about it. Let’s think about it. And let’s see if it’s likely to persist through the season.

Before we begin, I strongly recommend taking a few minutes to read this piece, which was published yesterday morning by my very talented colleagues at Baseball Prospectus (most notably Jonathan Judge). What the piece does, in its essence, is take a large volume of information that had, previously, been largely context-free (in this case, the exit velocity data provided by Statcast) and contextualize it to account for a wide range of complicating variables. It’s sophisticated stuff, and even if you have little interest in baseball research beyond its applications to the Cubs, I think you’ll get something out of it.

But anyway. Back to Bryant. One of the implications of the piece is that the way a batter typically strikes the baseball (what might loosely and carefully be called his “batted ball profile”) can be used to model—with a reasonable degree of accuracy—that batter’s total expected offensive contributions to his team. Thus, Judge and his colleagues have introduced a new statistic to the lexicon: Pred_Runs. They define the statistic as “the number of runs we would expect the batter/pitcher to generate/prevent based on the adjusted exit velocity and launch angle of the ball off the bat.”

Why is this interesting? Because, thanks to the magic of mixed modeling and lots and lots of previous work by the BP team, we already have data about how many runs a player has actually generated, at every point in the season. It’s called Act_Runs, and it’s defined as “the raw number of runs generated while the batter/pitcher was involved so far this season.” It, too, is a model—you won’t be able to count each run it accounts for in a box score, one for one—but, critically, it is descriptive rather than predictive. This means that by comparing its values, at any given point, to those of Pred_Runs, we can get a value that suggests at another value that looks rather like luck.

If that all sounds rather vague and wishy washy, I’m sorry. That’s the nature of throwing a caveat into every subordinate clause. Let’s try this again: people I work with wrote a Fancy Math Thing that guesses how many runs a player should have created, based on how hard they hit the ball. By comparing that number to the number of runs the players actually create, we can start to understand whether or not they’ve been unlucky in how they’ve hit the ball.

Is that better? I think so. Even if it’s not, let’s move on, because this is enough throat-clearing.

Point is, by subtracting Pred_Runs from Act_Runs, you can get a new statistic called (not particularly creatively) Pred_vs_Act. If a player was expected to produce more runs than he actually did, the value for the statistic is negative, and the player was (probably, unless the model is really wrong) unlucky. If the player was expected to produce fewer runs than he actually did, the value is positive and the player was lucky. And, right at the top of the Cubs’ leaderboards in the unlucky category? Kris Bryant. And that’s how we get to this point.

Based on how hard he’s hit the ball (93.2 miles per hour, after adjustments) and in which parks and under what circumstances, Pred_Runs predicts that Bryant would have produced about 12 runs of value so far this year, had he been neither lucky nor unlucky. However, Act_Runs records that Bryant has in fact only generated about 7.8 runs of value, meaning that he’s off (relative to his predicted performance) by 4.2 runs or so. Which, as it turns out, is a larger differential than that of any other Chicago Cub, and in fact one of the larger differentials in the game.

This is sort of interesting. If you are the type who likes to look at BABIP, you’re probably looking at Bryant’s .320 mark in that category and saying, well, this is a guy who’s likely to arrive a little bit closer to the league’s .300 average mark as time goes on. In other words, you’d think he’d been lucky, rather than unlucky. (If you were a little bit more sophisticated, you might point to Bryant’s .350-ish BABIPs throughout his career to date and reach the opposite conclusion, but that’s not the point here.) What Pred_vs_Act tells us, in contrast, is that Bryant has actually produced quite a few runs less than you would expect him to, given how hard he’s hit the ball.

Why is that? Well, I don’t know. But I have a guess. And here’s that guess: it’s because he hits the ball in the air a lot, and intentionally so. Bryant’s whole swing—though, to be fair, this is less true this year, as he’s flattened it—is built to generate fly balls, and lots of them. Once that’s accomplished, his inhuman strength conspires with the launch angle to create a lot of home runs. That’s the power you know and love. But the consequence of that is that on the balls Bryant just misses, even the ones he hits hard, the ball pops up into the air, or arcs through the sky in a lazy fly ball, in a way that’s especially easy to catch. Hard hit, but not that hard to turn into an out. This, I’m guessing, has created the combination of hard-hit balls and bad “luck” that you see in the data for this year and last (when he had a -11.8 Pred_vs_Act).

Now, some of this is already accounted for by the model, which includes within Pred_Runs terms for launch angle. However—and this is a very tentative however—I think it’s possible that players with Bryant’s particular profile (strong and with flyball tendencies) are overpredicted by the system. A quick glance over the list of players who the system similarly believes are unlucky shows a lot of folks rather like that: Chris Carter, David Ortiz, J.D. Martinez, Matt Kemp, Ryan Howard, etc. Something in these folks’ profile, I suspect, leads the system to believe that they should be performing a little bit better than they actually do.

So will the bad luck continue? Again, I don’t know. But if my hypothesis is correct, then probably so. In fact, if the hypothesis is correct, it’s not really reasonable to call it bad luck at all. Bryant’s “bad” Pred_vs_Act numbers are a natural consequence of his swing, which he’s designed to generate as many home runs and doubles as possible. So, maybe Kris Bryant isn’t that unlucky after all. Just a little thing to consider as you go through your day today and, especially, as you ponder the ball he nearly hit over the fence in Milwaukee last night, and which was caught instead by Kirk Nieuwenhuis.

After publication, sections of this piece were rewritten based on comments by Brad (@ballskwok). Thank you, Brad!

Lead photo courtesy Charles LeClaire—USA Today Sports.