If you haven’t already, go read David Laurila’s Q&A this morning with Dan Rosenheck, writer for the Economist and New York Times, who gave a presentation on predicting BABIP at the Sloan Conference last week. In that piece, Rosenheck notes that he created a model using just two variables — infield fly rate and rate of contact on strikes — that helped explain 15% of the variance in a pitcher’s future BABIP. The part about infield flies helping reduce BABIP has been noted before, as others have created takeoffs of ERA estimators that incorporate batted ball data — SIERA, tRA, bbFIP, etc… — and Steve Staude wrote a Community Blog post on this topic back in October, also identifying infield fly rate as a significant explanatory tool for BABIP. The potential explanatory effects of inducing popups and the link to Z-Contact% is fascinating, however, and makes Rosenheck’s study a real step forward.
It makes perfect sense that infield flies would help explain some of the variation in a pitcher’s BABIP, of course, since infield flies are almost always outs. In fact, in 2012, there were 4,377 batted balls that were categorized as infield flies in Major League Baseball, and only 13 of those went for base hits. Another 28 did not result in an out due to an error by the fielder, but even with 41 non-outs, that leaves IFFBs with an out rate of 99.1%.
Infield flies are, for all practical purposes, the same as a strikeout. They are basically an automatic out, runners do not advance on infield flies, and perhaps most importantly, we can state with a pretty high level of confidence that the relative abilities of the defenders have nothing to do with the outcome of the play. Sure, maybe you or I wouldn’t turn every IFFB into an out, but for players selected at the Major League level, there is no real differentiation in their ability to catch a pop fly.
So, based on those characteristics, an argument could actually be made that infield flies are essentially a fourth fielding independent outcome. No outcome is 100% fielding independent, of course, as catchers do have some ability to influence BB and K rates, and occasionally a HR is either robbed or knocked over the wall by an outfielder making a leaping grab on a long fly, but by and large, BB/K/HRs are mostly independent of the pitcher’s teammates, which is why they are the three variables in FIP. But, Dan’s comments got me thinking — if an infield fly has the same logistical outcome as a strikeout, should we just give a pitcher credit for IFFBs in the same way we give them credit for Ks?
In Tango’s bbFIP — which adjusts for all batted ball types — he does exactly that, adding Ks and IFFBs together and multiplying them by the same factor. It works because strikeouts and infield flies have almost identical run values. In 2012, a strikeout was worth -.265 runs, while an infield fly was worth -.268 runs. All the empirical data suggests that a pop up and a strikeout are essentially equally good for the pitcher, and both of them have little to do with the defensive support a pitcher gets from his teammates.
For predictive purposes, you definitely want to make a distinction between the two, as getting strikeouts are far more consistent from year to year than generating popups. Last year, Bill Petti ran the year-to-year correlations for basically any measure you can think of, and he got .82 for K% and .37 for IFFB%. There is no question that strikeout rate is more predictive of future strikeout rate than infield fly rate is of future infield fly rate.
However, just because FIP has been used to predict future ERA does not make FIP a predictive metric. It is a descriptive metric that happens to predict future events better than ERA, but as Glenn DuPaul wrote about extensively last year, if you wanted FIP to be predictive, you would use different weights for the formula than the ones that are currently in place. The weights for FIP come from the run values of the events being measured, not how well those events predict future events. In fact, home run rate has a y-t-y correlation of .42, much lower than either strikeout rate or walk rate, but it is included in FIP because it is a defensive independent outcome that the pitcher should be held responsible for.
Maybe we should consider that the IFFB is essentially not that different from a HR, in that it measures the results of specific batted balls that have a distinct run value and that aren’t influenced by the defenders behind any given pitcher. Just like we penalize pitchers for giving up home runs, logic would suggest that we should be giving them credit for infield flies.
At this point, all of the metrics designed to account for batted ball types have not just stopped with IFFB, but have also worked to apportion credit for GB%, OFFB%, and LD%. Knowing the results of those calculations can be useful, but those three batted ball types can all be described as having significant defensive contributions to the outcome, and thus, they don’t really belong in Fielding Independent Pitching. Infield flies, though, are a different animal from the other batted ball types, and I think you could make a pretty good case that they are essentially fielding independent. So, what happens if we just construct a very simple adjustment to FIP that treats IFFBs as Ks, and give them to the pitcher instead of the defenders?
It’s actually pretty easy to do, since the FIP formula is just basic math, and increasing the total number of Ks to include K+IFFB is just adding one additional term to the calculation. However, we do have to make an adjustment to the constant that allows FIP to equal league average ERA, since we’re increasing the number of outs that we’re crediting to the pitcher, which drives down FIP for each hurler. Last year, the constant for FIP was 3.095, but after including IFFBs in the K term, the new constant goes up to 3.196.
Just for fun, here’s a table of the 10 biggest gainers from FIP w/IFFB, or IFFIP, or whatever you want to call this slightly modified calculation.
In eight of the 10 cases, the pitcher beat their FIP, and in each of those eight cases, IFFIP is closer to the pitcher’s overall ERA than FIP was. While the group outperformed their FIP by 20 points, their IFFIP was almost exactly dead on to their overall ERA. That’s actually just kind of lucky, as infield fly rate doesn’t explain all of the difference between a pitcher’s FIP and his ERA — there’s also BABIP on non-IFFBs and the sequencing of when different events occur — but including infield flies does help to explain part of variance between a pitcher’s FIP and his ERA, and for pitchers who happen to have generated a lot of infield flies, it can really add up.
For the 88 qualified starting pitchers in Major League Baseball last year, the overall correlation between FIP and ERA doesn’t change much by including IFFBs with Ks. The three true outcome formula returns a .77 correlation between the two, while including IFFBs moves the correlation all the way up to .78. Please don’t take this post as evidence that FIP is completely worthless and that it has been debunked. That said, there are pitchers who got a lot of infield flies, and infield flies are basically guaranteed outs, and giving pitchers credit for those guaranteed outs does make some logical sense when using FIP to describe what happened in the past.
And, of course, FIP is the metric we use in calculating pitcher WAR here on FanGraphs. We use FIP as a descriptive metric that tells us what happened in the three categories that we’re pretty sure that the defense had little impact on, and that we can definitively ascribe to the pitcher. Perhaps it makes sense for us to consider using this kind of slightly adjusted FIP that also gives pitchers credit for their IFFBs when calculating pitcher WAR. It’s certainly not a decision we’re going to make without consideration, but it’s probably worth asking the question.
We’ll talk about this internally, I’m sure, but I’m also curious to know what you guys think – would you prefer that we adjusted pitcher WAR to give pitchers full credit for the infield flies they generate?