Oh, I hear you. This isn’t what normal people do when they’re discombobulated, not even a little bit. To that I say, you’re not wrong. Also, though, I’m not a normal person. This is my job, and daydreaming about work is inevitable, not weird. In any case, I came up with an incredible idea, a way to work out the next Paredes before teams did. I was a genius. Here’s the bad news, though: I don’t really remember it now that I’m lucid again.

That’s a bummer, but it’s OK, because in trying to reconstruct my thoughts, I think I came up with a pretty cool way of contextualizing how much it pays to sell out for pulled contact. As an added bonus, I got to pore over a ton of data and play with it to my heart’s content. That’s the dream, coming up with some silly junk stat in a haze and then spending hours manipulating data to show that it’s worthwhile.

This one’s going to be a little bit stream of consciousness, so let’s start where I started. I wanted to come up with a good proxy for how hard people hit the ball. There’s barrel rate, average exit velocity, hard hit rate, blast rate, whomps per whiff, hard-to-soft-contact ratio, and I only made up one of those stats. I went slightly off the beaten path and used Tom Tango’s “best speed,” the average exit velocity of the top half of their batted balls.

I used this partially because of the study I linked to, and partially because I wanted to work out how to calculate it using Python. First, I took every non-foul batted ball from 2021 through 2023. I used those end dates because I wanted to: a) pick years where Statcast used the same technology to capture batted ball events in each year, and b) throw 2020 out. From there, I grouped each batter-year separately, threw out the bottom half of values, and took an average of the top half. I also made a note of how many batted balls were in each batter-year sample. The code’s here if you want to goof around with it yourself.

That was useful in one way: It proved to me that best speed is a pretty good method of approximating damage on contact. For example, here are all those batter-seasons broken out into quartiles:

For another thing, if you run a linear regression, you get a whopping 0.44 r-squared for best speed, which means that 44% of the variation in wOBA on contact can be explained by variation in how hard the top half of a player’s contact is. That’s a really strong result, and in keeping with the Tango study that came up with the statistic. Anyway, now I had best speed for every player-season (min. 100 batted balls) over the last three years, and I could go from there.

I’m a big fan of chopping things up into quartiles, so I came up with another way to divide things: air pull rate. One frustration I’ve had in the past when working with batted ball direction stats is that pull rate isn’t as great of a measure as you’d hope. The issue: Batters pull grounders at a much higher rate than they pull balls hit in the air. When you’re ordering by pull rate, you’re also sneakily ordering by groundball rate, and that’s definitely not what I wanted to do here. So I only looked at pull rate on balls hit in the air. I took that value for each batter-year in my sample, as well as their production on all balls hit in the air.

That’s a lot of gibberish, I know, but I’m working toward a specific end goal. I want to know whether powerful hitters get the same tailwind from putting more pull in their game as mid-tier power guys, and whether very light hitters get any benefit at all from it. So I took those quartiles that I’d already produced and divided *them* up based on air pull rate. Each group had 303 hitters, so I made four groups of 75-76 (I went 75/76/76/76, for the record) in descending order of air pull rate. Here, for example, are the four quartiles of air pull rate for the hardest hitters in the game, along with the wOBA they produce when they elevate:

There doesn’t seem to be much there, huh? If you hit the ball hard, you tend to prosper, but it doesn’t appear that your spray tendencies matter much. The gap is only 20 points of wOBA, which isn’t huge – but that’s maybe what you’d expect for guys frequently cranking balls way over the wall. OK, fine, I’m not surprised by that. What about the next tier down, the hitters who have above average but not elite best speeds?

Huh. Really thought there’d be more of an effect there. Maybe these guys are still too powerful, though. Paredes, the patron saint of this talent, is in the bottom half of best speed. Let’s try those instead:

I’m really surprised by the lack of magnitude. Pulling fly balls is more valuable than hitting them elsewhere, particularly at lower exit velocities. And yet the hitters who do it more aren’t prospering as much as I’d expect. It’s not a matter of hidden exit velocity differences, either. I broke down each of these quartiles by exit velocity on balls in the air and there’s no difference at all.

Now, this doesn’t mean that pulling the ball is bad, or counterproductive. You can see in the numbers that pulling your medium contact is wildly valuable, in fact. But when taken in the aggregate, a hitter’s pull tendencies just don’t seem to matter – certainly, they matter far less than quality of contact. It’s not an issue of quartiles, either; I recast the data split into five, six, 10, whatever number of divisions, and nothing much changed. The correlation between pull rate simply isn’t that strong; the difference between being the pull-happiest and oppo-happiest hitter is worth only a handful of points of wOBA on your air contact – nearly nothing when it comes to your overall batting line.

That said, I didn’t come away with nothing. Since I had all this Baseball Savant data already, I decided to redo the analysis using xwOBA, which is a place where pulled batted balls seem to be underrated. Here, you can see an interesting effect:

The batters who pull the ball least have the highest xwOBA in every instance. This trend is repeated in all four quartiles. The reason for this is straightforward: Pulled batted balls consistently outperform their xwOBA, particularly at middle-to-low exit velocities. That’s a function of stadium geometry, but also of the mechanics of hitting a ball; pulled batted balls and oppo batted balls might have the same exit velocity and launch angle, but plenty of their other characteristics differ based on how the contact was produced. The average exit velocities are the same in all of these groups, and yet oppo-heavy contact does better by xwOBA, which likely means that it’s hit at more “optimal” launch angles that nevertheless produce equally as well as the pulled contact with worse xwOBA numbers.

Is this meandering? You bet. As I’ve no doubt mentioned in a lot of my articles recently, it’s the middle of winter and you’re reading about baseball, so that comes with the territory. My takeaway is that best speed is a really good predictor of how much damage hitters do when they put the ball in the air. Within that, looking for guys who pull their contact most frequently doesn’t add as much as I’d hoped.

There’s surely more research to do, don’t get me wrong. My next investigation is going to be into what happens to hitters who change their pull rates, which seems like a fruitful territory to investigate. For now, I think the main takeaway is that best speed and some measure of air/ground tendencies – GB/FB ratio, perhaps – do a really good job of predicting wOBA, particularly wOBA on balls hit in the air. Adding pull rate to that mix doesn’t seem to help much, but adding xwOBA doesn’t seem to help much either, because it likes the opposite field contact too much. There’s undoubtedly a code to crack somewhere in the data, and maybe I’ll get there someday, but for now I’m happy to say that this investigation has reached a stopping place, and I can rest.

Source