March 15, 2006
Lies, Damned Lies
Another Look at Plate Discipline
Juan Pierre. Manny Ramirez. Which player has better plate discipline? Ramirez has drawn an unintentional walk in 11.3% of his plate appearances during his major league career, and Pierre 5.9%. For most of us, the discussion begins and ends right there.
Now let's say that you're a pitcher. Juan Pierre. Manny Ramirez. Which player are you going to throw strikes to?
Kevin Goldstein's excellent article reminded me of some research I've done on this subject, which got lost in the shuffle of PECOTA and Baseball Between the Numbers and everything else. Plate discipline is a skill. There's some room for debate about whether it belongs under the broader umbrella of "hitting," or whether it ought to be its own category, but either way it's a skill. The problem is that unlike skills such as power or speed, which are well reflected in home run and stolen base totals, there is no clean, commonly-available statistical measure of plate discipline. As I'll (hopefully) be able to persuade you, walk rate tells part of the story, but not the whole story. Looking at strikeout rate in addition to walk rate gives us a more robust view of a player's hitting approach, but still doesn't get to the core of the plate discipline issue.
I think of plate discipline as consisting of three, somewhat interrelated abilities:
"Managing the at bat" is something that I hope to study in more detail over the course of this season, complete with some inroads into game theory and the like. Basically, it covers things like whether a hitter should swing at a marginal pitch on a certain count. You're a fastball hitter, the count is 0-1, and the pitcher throws you a slider that looks like it just might be hanging over the inside portion of the plate. Swing? Or wait for something better, at the risk of falling down by two strikes? This is a topic that's been largely underexplored and we'll leave it alone for now, although Tom Tango, Mitchel Lichtman and Andy Dolphin make some excellent strides in The Book.
But back to this research that I mentioned before. During last year's postseason, ESPN.com published Inside Edge scouting reports for each of the eight playoff teams. Included in these scouting reports was "chase percentage," the frequency with which a hitter swings at a pitch outside of the strike zone. Chase percentage comes much closer than walk rate or strikeout rate to measuring "true" plate discipline ability as we described it above.
Because the Inside Edge information was behind ESPN.com's subscriber wall, I'm not going to report individual figures. I also had to be a bit creative in interpreting the data. The Inside Edge reports included chase percentage for each of the eight "corners" of the strike zone individually, but not an aggregate figure; I simply averaged the eight individual figures to get an overall number. The data was also missing or incomplete in a few cases. But I was able to estimate chase percentage for 71 hitters on playoff clubs who had at least 250 regular season PAs. The average chase percentage for these players was 28%, with a range of between 13% and 45%.
Let's start by looking at a couple of simple correlations:
Chase Percentage / Unintentional Walk Rate: -.68 Chase Percentage / Strikeout Rate: .09Unsurprisingly, there is a pretty large inverse correlation between chase percentage and walk rate. However, there isn't much of a relationship between chase percentage and strikeout rate. Why not?
I suspect this is because certain types of hitters have more of an incentive to chase pitches--specifically, good hitters who get fewer good pitches to hit. If you're Manny Ramirez, you swing at that 2-0 pitch that just might drop off the table, but could wind up in the right field bleachers if it doesn't. If you're Juan Pierre, you take the sucker and try and draw a walk, since the best you're going to do is wind up on first base anyway. Here's another correlation:
Slugging Percentage / Unintentional Walk Rate: .38One stathead cliché is that walks produce power. Although there is an important relationship between the two, I believe it is actually the other way around: power produces walks. In the very rigorous, multivariate regressions that I've done as part of PECOTA, I have found that isolated power and batting average are both favorable predictors of walk rate, but the reverse isn't true: walk rate doesn't predict power.
This conclusion ought to be reasonably intuitive, and gets back to the point that we made before: major league pitchers are smart enough to throw many more hittable pitches to the Juan Pierres of the world than the Manny Ramirezes. We can see this more clearly by throwing chase percentage, slugging percentage and walk rate into a regression equation of their own:
Unintentional Walk Rate = Chase x SLG x (-.77) + SLG * .39 (R^2=.65)That's probably not very interesting to you in the abstract. Here are the predicted walk rates for players with different chase percentages, holding SLG constant.
A power hitter that chases a lot of pitches will often draw as many walks as a light bat that almost never chases, simply because the former is being pitched around so much more frequently; Pierre might well have better plate discipline than Ramirez. We can also look at the relationship between slugging percentage and walk rate, holding chase percentage constant:
Note that, if we extended these three lines leftward to include a SLG of zero, we'd find that the player wasn't predicted to draw any walks at all; he'd simply never see a pitch outside of the strike zone unless the pitcher was under duress. Walks aren't always "mistakes" by pitchers so much as they are part of a risk-minimizing strategy that involves sometimes giving up a little (placing a runner on first base) rather than a lot (placing a couple of runs on the board). Of course, the best pitchers will be able to have their cake and eat it too, by throwing tough pitches within the strike zone, and the best hitters will be able to force the issue by not chasing a wayward pitch.
So, we should be giving a little bit more credit to the Pierres and Scott Podsedniks of the world; it's awfully hard for them to post even the modest walk rates that they do. On the other hand, this doesn't bode well for their futures. Podsednik's walk rate before the All-Star break last year was 10.7%. After the break, once it had become clear that he wasn't capable of hitting for any kind of home run power at all, his walk rate halved to 5.4%. The scouts have it right in one respect: all the plate discipline in the world is useless if it isn't part of a good, well-rounded hitting approach.