Analyzing Swing Tendencies
A look into swing rates, zone swing rates, chase rates, and the implications they have hitter performance.
Plate discipline is a key grouping of statistics that a lot of people like to talk about. How often does a hitter swing? How often do they swing at bad pitches? At good pitches? What does all of this mean for their actual production?
To me, it seems like a lot of people talk a lot about these statistics but don’t really understand the context or the implications of them. So I’ve set off to learn a bit more about these statistics myself and share some of the things we can find out about them.
The Problem with Chase Rates
One stat analysts love to cite is the chase rate - the percentage of pitches out of the strike zone that a hitter swings at. It’s otherwise known as O-Swing% (the O standing for out of the zone)
For context:
League average: 28%
Worst in Recent History: 51% (Francisco Mejia 2022)
Best in Recent History: 12% (Juan Soto 2021)
So a low chase rate is good, and a high chase rate is bad. However, the one thing people fail to realize, or at least fail to mention, is that overall swing rate is the main driver of chase rate. Most often, the best chase rates are coming from players that just don’t swing at anything at all. I could go up to the plate and not swing once all year, and I’d have an ELITE chase rate, but I’d also strike out about 85% of the time and never get a hit.
Here’s the relationship of Chase% to Swing%:
R^2 = 0.75
That means that swing rate explains 75% of the variance in chase rate. In other words, you can predict a hitter’s chase rate quite accurately by just looking at their overall swing rate. The closer an R^2 gets to 1, the more the input stat (swing rate in this case) predicts the output stat (chase rate in this case).
This is important to note because it really takes away the purpose of citing a chase rate. Juan Soto has an elite chase rate right now, and while the guy clearly does have a great eye at the plate - the low chase rate is driven by the low overall swing rate. It doesn’t mean much to say he has a great chase rate.
Swing Rate Correlations
A quick sidebar here about swing rates. Despite some very notable people saying otherwise, swing rate doesn’t tell us anything about a hitter’s strikeout rate:
R^2 = 0.001
The two things are not related at all.
Swing rate does do quite a bit of work to predict a hitter’s walk rate though:
R^2 = .56
This makes intuitive sense. You need to take four balls to take a walk, you cannot take a ball if you swing, therefore - if you stop swinging, you’ll walk more.
You might think the inverse of this would be true with strikeout rate - the more you swing, the fewer called strikes you’ll take and the more balls you’ll put in play, etc. - but you’d be wrong in thinking that. If you make contact at a very high rate, you don’t need to swing very much to put a ball in play.
Let’s take a quick example of how we can use this. Colton Cowser just recently debuted for the Orioles. At the time of this writing, he had logged just 18 PAs. That’s not enough data for us to know hardly anything at all about him, however, swing rates stabilize pretty quickly - so we can look at his low 41% swing rate and guess that he’ll be a guy that walks more than average. We can back this up by seeing that his swing rate in the minors was below 40%, so it’s a safe bet that he will be a high walk rate player.
Enter Swing Decision%
So we can’t learn much from chase rate by itself. What we want to know is to consider multiple things at once. We like to get things into one number as best as we can, so one possible solution to this is to look at the percent of a hitter’s total swings that are at pitches in the strike zone. I call this Swing Decision% on my dashboard. It’s not a great name, and it’s used by some other people for some way more complex stuff, so maybe I should rename it.
League average SwingDec%: 74%
Anyways, it’s not the perfect stat - but it’s much more useful than looking at just chase rate, or taking chase rate and in-zone swing rate together and trying to do some the math in your head, etc.
Here’s a look at it, presented by plotting Z-Swing% (in-zone swing percentage) against O-Swing% (Chase Rate) and then coloring the dots by this Swing Dec% metric.
You can see that the blue dots are to the left, meaning the good swing decision hitters are hitters that are not chasing. The dots also get bluer as you move toward the top-left of the plot. Those dots would be the players that swing at a lot of the pitches in the zone but very few of the pitches out of the zone.
You can see that no player really pushes close toward the possible extremes (top-left or bottom-right). The perfect hitter would have a 0% chase rate and a pretty high Z-Swing% (it wouldn’t be 100% because not all strikes are worthy of a swing - sometimes the pitcher just gives you nothing to work with). But the best examples we can find are these players:
2021 Brandon Nimmo (67% Z-Swing%, 15% O-Swing%)
2022 Austin Barnes (66% Z-Swing%, 15% O-Swing%)
2021 Juan Soto (60% Z-Swing%, 12% O-Swing%)
2021 Brett Phillips (65% Z-Swing%, 17% O-Swing%)
2023 Taylor Walls (66% Z-Swing%, 18% O-Swing%)
So three of these guys aren’t very good hitters, which doesn’t bode well for the usefulness of Swing Decision% as a stat.
When we look at the names at the bottom of the list:
2022 Javier Baez (68% Z-Swing%, 48% O-Swing%)
2021 Javier Baez (76% Z-Swing%, 45% O-Swing%)
2023 Javier Baez (70% Z-Swing%, 46% O-Swing%)
You don’t have to be a baseball stats guy for long to know that Baez is the least-disciplined hitter in the league, so it’s a point in favor of Swing Decision% here. This stat can and does pick out very undisciplined hitters, and that stuff hurts you overall.
Swing Decision% Relationships
We know that Swing% predicts your Chase%, your Z-Swing%, and your BB%. The next question is - does Swing Decision% predict anything more useful for us? Well, it does predict your BB% a little bit:
R^2 = .37
But it does significantly worse than Swing% by itself, so it’s not useful in that regard - just use Swing% if you’re trying to predict BB%.
It does nothing to tell us about K%:
R^2 = 0
It does diddly to predict your OPS:
R^2 = 0.007
As for xwOBA - you guess it, no help at all:
R^2 = .036
You could say that the R^2 for xwOBA is higher than for OPS and K% and that the trendline is a little bit upward, so there’s a little tiny bit of signal there. I suppose I could grant you that, but it’s close enough to zero that it’s really not worth noting.
You have plenty of great swing-decision hitters with a terrible xwOBA. That’s because xwOBA is driven by launch speed and launch angle, and your decision on what pitches to swing at tells us nothing about what those numbers will be.
The one thing we don’t find in the plot is hitters with horrible swing decisions putting up high xwOBAs. All of the xwOBAs above .350 (save two seasons from Albert Pujols and Salvador Perez) had SwingDec% above 65%.
So that’s a possible takeaway here. If you’re not swinging at the right pitches, you’re probably not going to impact the ball well over a full season’s sample.
That’s bad news for these young hitters struggling with this in 2023:
Elehuris Montero: 51%
Oscar Colas: 55%
Yainer Diaz: 58%
Elly De La Cruz: 59%
Mickey Moniak: 60%
Casey Schmitt: 60%
Is It Sticky?
The next question is - does this year’s SwingDec% tell us anything about next year? The answer is yes!
Sorry to switch up the freaking scatter plot design on you there, I had to go to Excel for that bad boy. Correlation there is .79, a very strong relationship.
This is not to suggest that guys like Colas and Elly are going to stay bad at this next year. We’re looking at small samples with them, which weakens the numbers, and it’s also certainly a fact that young players will see wider variances year-to-year in stats like these since they are currently becoming the Major League hitter they will eventually be. There’s an adjustment period, and it’s smart to expect young hitters to swing at fewer bad pitches as they see more of what a bad pitch looks like at this level.
So no reason to panic about those names, but if you have a player consistently putting up bad SwingDec%, it’s unreasonable to think he’ll change next year.
Free Conclusion
I am going to slap up a paywall here soon and give more information to the paid subs, but I wanted to wrap this up for the hoi polloi appropriately.
Chase Rate is a weak stat to look at by itself, stop doing that
Swing Decision% is much better, however, even that stat doesn’t tell you much about what actual performance will be besides walk rates.
So this is a long post that comes to no conclusion other than that there is no conclusion. But this is still good to know. It’s just as important to know how NOT to look at a stat and which stats to NOT look at as it is to know which stats to look at and how to look at them.
There’s a lot of discussion about swing decisions and plate discipline, so I hope this can serve as a helpful resource to learn more about it for people in the future.
Now we’ll hit a paywall and I’ll give some more leaders and losers of the 2023 season, and I’ll give the link to the analysis dashboard and dataset that I used for all of the above.