Are Popups a “Skill”?
In light of yesterday’s article on Infield Fly Balls and xFIP, there were some questions and debate about if popups are something under a pitcher’s control, or what you might call a skill. After reading the comments, I was somewhat doubting that my research might not have been thorough enough to essentially rule them out as a skill (which is what I more or less did).
If you look at popups per ball in play on a year-to-year basis, you get a correlation of about .52, which would highly suggest that there is some “skill” in inducing popups. However, there is a very strong positive correlation between popups and outfield-fly-balls (.64), and a very strong negative correlation between popups and groundballs (-.72).
In other words, as outfield-fly-balls increase, so do popups. As groundballs decrease, popups increase. For comparisons sake, line-drives have absolutely no correlation with popups.
Since the correlations are so high, you can basically come up with an expected popup rate based on a player’s groundball percentage. To me, it actually looks non-linear:
So it seems that each player has a dynamic expected popup rate based on his groundball percentage. Now the real question is, do players popup rates diverge from their expected popup rates consistently on a year to year basis?
If you look at the above chart, you’ll see that there’s not much consistency from year-to-year. The correlation is about .18, which pretty much agrees with Mitchel Lichtman’s findings of .14 as quoted in David Gassko’s Batted Ball DIPS article. For comparisons sake, BABIP has a year-to-year correlation of .15.
So what does this mean for popups as a “skill”? I’d say they are sort of a skill that is closely tied to groundball percentage, but from the findings above, that’s about as far as I’d go. While there may be certain pitchers that prove to be exceptions (just as there are exceptions with BABIP), popups in general do not seem to be much of an independent skill.
All batted ball data is from Baseball Info Solutions from 2006-2009 for pitchers with at least 100 innings pitched.
David Appelman is the creator of FanGraphs.
Hey Dave,
What’s the regression function (I think that’s the right term) you came up with for the first graph?
Thanks!
It will begin to look a lot less nonlinear if you compare popups as a percentage of non-GB batted balls to GB% instead of just using PU%.
I did some analysis and I can’t say I’m entirely convinced popups aren’t a skill of their own, or at least aren’t influenced by something other than GB%
First of all, I know LDs are probably not a pitcher skill. I don’t know what type they “should” have been, so I’ll ignore them. So, my sample will consist of all pitcher seasons 2003-2009 with 100 or more non-LD batted balls (MLBAM/Retrosheet classifications).
Now, I know that naturally GB% increasing will reduce the other batted-ball ratios, so I will compare PUs as a percentage of PU+FB to GBs as a percentage of non-LD batted balls.
This gives me an R^2 of 0.87, which seems highly significant (regression equation: PU/(FB+PU) = -0.209*(GB/(GB+FB+PU)) + 0.329).
But, I then checked this estimate against the data. On average, the estimate was off by 21%. This doesn’t seem entirely convincing.
You might be interested in this, from six years ago! (I’ve been at this way too long).
http://www.hardballtimes.com/main/article/never-swat-an-infield-fly/
By the way, my “go-to” reference for batted ball analysis is the Gassko article in the 2007 THT Annual (following the Gassko/Bradbury article in the 2006 Annual). To quote from there:
“…pitchers not only have the ability to force infield flies…but that ability is partially independent of their ability to force fly balls overall.”
I’m sure there are significant differences in approach between their analysis and yours. For their original study (in the 2006 Annual), they included pitchers who had faced at least 350 batters in consecutive seasons. Including relief pitchers would have an effect on the results, because it’s been shown that elite relievers have a more consistent impact on batted ball type than starters.
BTW, you can read the 2007 Annual online for free here:
http://www.wowio.com/users/product.asp?BookId=511
Another thing to consider, is that IFFB% is a park factor as well, I think it’s the A’s stadium that inflates IFFB% considerably, due to large amounts of foul ball territory.
Thanks for the article. Interesting what the data can reveal.
2 Qs though:
On this article – does your finding necessarily show that no individual pitcher’s ability to induce popup is more than luck, notwithstanding their ability to induce grounders? I seem to remember a legend about maddux’s ability to “make” the batter popup to the left or right side of the infield, as the situation dictated.
Also, this article makes me wonder – are foul balls similarly the product of luck, or pitcher’s skill? do better pitchers induce more foul balls?
A reason pop-up rate may go up with flyball rate is because in order to give up that many flyballs in the major leagues, an extreme flyball pitcher will often to have a BABIP limiting skill to say in the major leagues, often in the form of a big pop-up rate. I’m sure this has already been done, but if you looked at extreme flyball pitchers, I’m sure you find that on average they have some BABIP limiting skill (possibley HF/FB reducing skill).
Isn’t it more likely that extreme flyball pitchers stay in the league because of their strikeout and walk skills, limiting the impact BABIP fluctuations can have?
I’m not saying that there isn’t a BABIP-limiting skill that some FB pitchers might possess (such as “Playing for the Padres”), but it seems more likely to me that they simply must do well in the other two true outcomes to remain effective.
Uhh… Lesseee…
If a pitcher had a whole in his batted ball game, what would he do to make more outs?
A. Create a Batted Ball skill to get more outs.
B. Not Walk a Lot of Batters, so that batted balls have a lower run value.
C. Create outs himself, using strike 3 skills.
I’m pretty sure that, on average, you find a lot of B and C in flyballers, and that A remains something we’ve never discovered with any great certainty.
Good point. David showed that home run rate per outfield fly goes down as fly ball rate goes up. Also, flyball pitchers tend to increase their strikeout rate, too.
edit: perhaps not on average actually, but a larger amount that other subsets of pitchers, and I mean besides the advantage allowing few groundballs helps your BABIP
I wonder how much the pitch type affects pop up chance. I think someone on your other article gave the example of Rivera as an outlier, I wonder if maybe a cutter is a popup inducing pitch.
I wonder if there is any correlation to swings and misses on fastballs, especially those up in the zone. It is not surprising that there was no correlation on line drives as the position of the ball on the bat for a pop-up is fairly far away, as well as the opposite effect for ground balls as the position of the ball when it meets the bat is even further away. I would not be surprised to find out that those with a greater percentage of pop-ups also generate a greater percentage of swings and misses on fastballs.
And you wouldn’t need to guess, it’s common knowledge that the whiff rate on fastballs increases the higher it goes, and flyball pitcher’s work up in the zone because it’s easy to coerce hitters to get under the ball. I don’t care for the infield fly stat unless it’s consistently well off the norm, because the only difference between a flyball into very shallow left and a popup to the shortstop is a couple of feet, yet they get different classifications.
If memory serves correctly, infield pops don’t occur more often on high fastballs.
To go along with what Yankee suggests:
Part of the reason that Popups may have an inverse correlation to ground balls could be due to a selection bias.
Lets assume that groundball pitchers are generally more successful at avoiding runs then flyball pitchers, all other factors being equal.
I think its also fairly safe to assume that clubs, especially prior to the last few years, track runs allowed results more then stats like groundball percentage to select who translates to the majors. If you don’t allow runs, you get to the show.
If these assumptions hold, then most pitchers who do not get ground ball outs must have another skill to fall back on. For most flyball pitchers, that skill is striking out hitters. There are a few flyballers, however, that do not strike out many hitters.
I suggest that these flyball pitchers who don’t strike anyone out have a higher ability to induce popups. Now, this skill may be harder to measure then groundball rates and strikeout rates but there does seem to be some sort of a trend.
This would explain why there is a negative correlation between ground balls and popups (besides the fact that GB pitchers tend to work lower in the zone) that popup pitchers make it to the majors primarily on this skill while other fly ball pitchers use the K instead.
Now this is all just off the top of my head so I’d need to crunch some numbers to figure out if the data supports any of what I just babbled. Hope this makes some degree of sense.
It’s possible it is a skill, but as I said in the other thread, it could also just be luck.
I think that maybe we’ve fallen for the two big problems with using correlation coefficients. For one, non-gb pitchers have to get by on K/BB ratio. This implies they can get Ks. I think Ks and pop ups share a cause, one really good pitch (be it a changeup or fastball). So pop ups aren’t really a skill, their just an attendant circumstance of a skill pitchers have to have to survive the selection bias of being successful without a great gb%.
Edit: I’d suggest that if anybody wants too check they’d do well to see if there is a strong correlation to pitch type linear weights (looking for pitchers with strong pitches not normally associated with causing grounders) and iffb%.
How about IF/FLY what is the y-t-y correlation for that?
I get 0.23.
Although, as I increase the minimum number of non-LD batted balls to 200 and 300, it becomes 0.27 and 0.38 respectively. It’s definitely repeatable.
How about IF/FLY what is the y-t-y correlation for that?
Let’s step away from the numbers for a moment and think about this philosophically and logically.
I’ve always like the definition of “skill” that says you have to be able to intentionally fail for it to be a skill. (The usual example is that a good chess player will be able to lose as quickly and efficiently as he wins, while someone trying to lose at roulette could just as easily win.) In that sense, is batted ball type a pitcher skill?
Clearly, pitches themselves are skills. If a pitcher has an awesome sinker, he can choose whether or not to throw it awesomely or as a fat floater over the plate. Sinkers are skills. Sinkers also might tend to induce ground balls because the batter swings on a normal plane while the ball drops, causing him to top the ball more often than he hits it squarely (unless, of course, he’s excellent at spotting and hitting sinkers, which would be a skill).
Taking the notion that some pitch types lend themselves to some batted ball results, it’s tempting to make the logical leap that pitcher skill equates to batted ball result. But it’s more honest to say that *batter* skill equates to batted ball result, and the inherent traits of pitch types are what the batter must overcome. A pitcher may be more skilled at throwing a sinker, but whether that sinker becomes a ground ball or a line drive is entirely up to the batter.
I’d like to see a measure of pitch types vs. batted balls (blind to who’s pitching them), then compare those baseline results to batter outcomes vs. pitch types (to see what batters are most skilled at hitting certain pitch types) and only then, given a good enough sample size, break that batter’s results down by pitcher, to see which pitcher has the most “skill” at throwing that pitch type.
To summarize my thinking: (some) pitch types have batted ball traits, hitting pitch types is a skill, pitching those pitch types is a skill, pitching to batted ball traits is therefore an echo of the amalgamation of these skill levels.