April 11, 2016

Rubbing Mud

Kill the Error

by Matthew Trueblood

Printer-friendly

Contact Author

Enrique Hernandez worked hard for first base in the seventh inning in San Diego on Tuesday. Though there were two outs and nobody on, and although the Dodgers led 3-0, and although Hernandez was facing Kevin Quackenbush (a right-handed reliever, which is usually enough to neutralize Hernandez), and although Hernandez was just coming off the bench to take a low-leverage plate appearance in the pitcher’s spot, he hung in there. He fell behind 0-2, but then took a pair of tough pitches, fouled off a good fastball on the outside corner, fouled off a good curveball that ended up just below his knees. On a 2-2 count and the seventh pitch of the plate appearance, Quackenbush left a fastball over the middle of the plate (though it was above Hernandez’s letters), and Hernandez hit it hard. Specifically, he hit it 102.8 miles per hour, according to Baseball Savant. Unfortunately, it was right at shortstop Alexei Ramirez. Ramirez gathered in the sizzling topspin grounder, set himself… and threw low to first base. Wil Myers couldn’t pick the short hop, and Hernandez had his base.

Of course, Hernandez won’t get credit for that anywhere. The play was a textbook throwing error, and it was called that way, so despite his tough at-bat and very hard contact, Hernandez got an 0-for-1 on the night.

Why isn’t the error dead by now? We’re out of good reasons. No longer is there any sense in using the statistic. Where once we might have needed the subjective calibration of the record on the part of an official scorer, we now have hard data on the majority of batted balls across the league. Even if it’s ultimately a defender’s bad throw or rock-hard hands that leads to a batter reaching base, we ought to acknowledge that hitting the ball hard (indeed, putting it in play at all) applied pressure and created an opportunity to profit from the defense’s suboptimal response to that pressure.

Using the Baseball Savant PITCHf/x search, I looked at all regular-season games from July 1, 2015 through Friday night, seeking groundballs hit between 102 and 103 miles per hour, by right-handed batters, with the bases empty. (Handedness matters, both because of the different ways shifts can affect left- and right-handed hitters, and because of where grounders are likely to be hit by batters of each type. Baserunner context matters, too. The same ball hit with runners on base is a different proposition, both more likely to get through the infield (because of the way defenses must align themselves with runners on) and much more costly if it doesn’t, since it might well be a double play in that situation. It’s better to just wash those differences out of the sample we’re examining.) I found 413 such batted balls, and here’s how they broke down:

· 147 singles

· 18 doubles

· 1 triple

· 1 fan interference (essentially, a double)

· 9 reaches on error

· 237 groundouts

In other words, if you take the traditional tack to scoring these batted balls, batters hit .404 when they made similar contact to what Hernandez did on Tuesday night. If you count errors as hits, that number rises to .426. Either figure is telling, since (depending on which data set you use) the league hit between .236 and .245 on all groundballs last year. Not all groundballs of similar speed are created equal, in terms of placement, in terms of trajectory, or in terms of spin, but this is a substantial sampling of balls hit about as hard as Hernandez’s was, and it suggests that he gave himself a 40-odd percent chance of reaching base by making the contact he made. In that light, doesn’t it seem strange to strip away credit simply because the way he reached base wasn’t the way many of the others who hit similar balls reached base?

There’s a problem with this logic. The implied suggestion, that Hernandez ought to have been given a hit, isn’t a perfect use of this data, either. In fact, it’s a downright unfair use of the data. It might be worse than the alternative. After all, if Hernandez hitting the ball the way he did gave him less than half a hit’s worth of a chance to reach base, giving him no credit for the plate appearance is a closer approximation of the right apportionment than is giving him full credit for a hit. No, what we need to do is to start giving partial credit for hits on all batted balls. This is the future—defense- and luck-adjusted offensive stats. Every batter who hits a ball the way Hernandez did gets 42 percent of a hit. Every left-handed batter who hits a fly ball to right field at between 100 and 101 miles per hour gets 66 percent of a hit. And obviously, there will be adjustments to the way we put batted balls into bins, by location, by launch angle, and there will be corrections for ballpark and weather, and instead of simply crediting batters with fractions of a hit, we’ll credit them with fractions of weighted runs created according to the matrix of possible outcomes suggested by their hits.

Yes, it’s a ways off. We’re not ready to rely on the samples we have to set baseline expectations for various types of batted balls. We’re not yet able to tell whether certain types of hitters have certain similar characteristics to their batted balls that make them better bellwethers for one another than for others. There’s still so much about batted-ball data collected via Statcast that we can’t, and don’t, know. I guess what I’m saying is, until the glorious future comes and we’re free from the tyranny of outcomes and can just judge batters’ and pitchers’ and fielders’ processes, can we please, pretty please, get rid of the damn error? Because it’s a relic of another time, and the data we do know how to use and process makes it look cruel and senseless, and we have enough cruel, senseless relics of another time roaming the Earth right now, you know?

Matthew Trueblood is an author of Baseball Prospectus. Follow @MATrueblood
Click here to see Matthew's other articles. You can contact Matthew by clicking here