It’s Time for a Pitcher to Throw 80% Breaking Balls
A few years back, I was sure that throwing too many breaking balls was bad for pitchers’ arm health. I wasn’t alone — there was some decent research backing up that hypothesis. As the methods for examining the question have become more refined, however, and further work has been conducted on the matter, it looks like we’ve found that it’s not so much breaking balls as velocity that most directly affects arm health.
Perhaps teams have been on the same journey, because curveball usage — and breaking-ball usage, in general — is up to heights we haven’t seen before.
| Season | Four-Seam | Changeups | Two-Seam | Breakers |
|---|---|---|---|---|
| 2010 | 40% | 13% | 18% | 29% |
| 2011 | 38% | 12% | 19% | 30% |
| 2012 | 35% | 12% | 22% | 31% |
| 2013 | 36% | 12% | 22% | 31% |
| 2014 | 35% | 12% | 22% | 31% |
| 2015 | 37% | 12% | 21% | 30% |
| 2016 | 36% | 12% | 20% | 31% |
| 2017 | 36% | 13% | 19% | 32% |
Changeups = changeups plus splitters
Breakers = sliders, cutters, curves, knuckle curves, and eephi
Breaking-ball usage has increased. That said, the uptick has been slow and gradual. Perhaps too slow and gradual. Maybe we should be pushing it.
We’ve written a lot about the juiced-ball theory, but when Ben Lindbergh recently published results from Major League Baseball’s internal ball testing — which produced no indication that the ball would be responsible for the marked increase in home runs — we were left scratching our head as to what the actual cause might be. Russell Carleton over at Baseball Prospectus thought that we could explain a good deal of that increase with a simple look at hitters’ tendencies. Take a look at what he found:
Starting in 2015, around the time of the current power spike, hitters seemed to be sitting fastball more. And why not? Even though fastball usage has actually declined among pitchers, it still accounts for more than half of all pitches. As a batter, you know you’re eventually going to get one. So, why not wait for one and swing really hard? You might miss and eventually you might strike out, but if you do, so what?
Within that piece, there’s evidence that batters are swinging less often and have pushed their average pitches per plate appearance close to four (3.8 last year), up from 3.5 earlier this century. What Carleton proposes is that hitters, now armed with the terminology and technology to discuss and improve how hard they hit the ball in the air, are sitting on fastballs because they’ll still get, on average, one or two per plate appearance.
What if they didn’t, though? We’re talking about a potentially healthier approach for a pitcher — fewer high-velocity fastballs — and also pushing the expected number of fastballs per average plate appearance under one. Can batters use the same approach if they might only see one fastball a night? Or two? The evidence is that velocity differential is not as important for whiffs on breaking balls — it’s not quite like a changeup in that regard, the breaking ball may not need the fastball as much as a changeup does.
What if a pitcher started throwing 80% breaking balls?
One possible outcome might be that said pitcher would walk a lot of people. And that’s a decent hypothesis. Here’s how those big groups of pitches from above rank in terms of zone percentage according to a piece by Harry Pavlidis from back the day. (Note: cutters have been removed from the breaking-ball group in this case.)
| Pitch | In Wide Zone % |
|---|---|
| Changeup | 43.6% |
| Breaking Balls | 47.8% |
| Four-Seamers | 56.0% |
| Two-Seamers | 54.2% |
Changeups = changeups plus splitters
Breakers = sliders,, curves, knuckle curves, and eephi
From Harry Pavlidis’ Pitch Type Benchmarks piece
The average breaking pitch doesn’t land in the zone as much as the average fastball, no. But if you had a pitcher with multiple distinct breaking balls — and good command of those pitches — you might want to consider pushing that breaking-ball percentage as high as it can go, right? Here are a few pitchers who throw a ton of breaking balls already and also feature good edge percentages on those breaking balls. They could be our best test cases.
| Pitcher | BBFC% | BB% | BB Edge% |
|---|---|---|---|
| Clayton Kershaw | 51.8% | 51.8% | 20.2% |
| Yovani Gallardo | 40.8% | 40.8% | 19.3% |
| Bronson Arroyo | 53.4% | 48.8% | 18.8% |
| Chris Archer | 46.5% | 46.5% | 18.4% |
| Andrew Triggs | 53.7% | 53.7% | 18.3% |
| Miguel Gonzalez | 41.9% | 41.9% | 17.9% |
| Jerad Eickhoff | 45.9% | 45.9% | 17.3% |
| JC Ramirez | 57.5% | 57.5% | 17.1% |
| Lance McCullers | 45.5% | 45.5% | 17.0% |
| Kenta Maeda | 49.0% | 44.2% | 16.4% |
| Drew Pomeranz | 47.7% | 45.0% | 16.0% |
| Jordan Zimmermann | 39.6% | 39.6% | 15.9% |
| Michael Pineda | 39.0% | 39.0% | 15.5% |
| Derek Holland | 43.0% | 43.0% | 15.4% |
| Jason Hammel | 37.6% | 37.6% | 15.2% |
| Jhoulys Chacin | 39.7% | 39.2% | 14.4% |
| Tom Koehler | 42.7% | 42.7% | 14.1% |
| Ricky Nolasco | 41.8% | 41.8% | 13.0% |
| Robbie Ray | 41.9% | 41.9% | 12.0% |
| Zack Greinke | 37.4% | 37.4% | 12.0% |
BBFC% = Above, plus cutters
BB Edge% = Breaking pitches within a baseball ‘s width of the edge of the strike zone divided by all breaking pitches.
Given his success, Clayton Kershaw probably doesn’t need to change anything up, despite his excellent command of his breaking balls. But Yovani Gallardo? Bronson Arroyo? Kenta Maeda? Ricky Nolasco? Jordan Zimmermann? Over the last few weeks, you could add Adam Wainwright and Josh Tomlin, and that would make sense. There’s enough struggle in their recent histories and enough upside in this idea that they might be inclined to go for it. They’ve each got multiple breaking balls and above-average command of those breaking balls — by this incomplete measure at least.
Of course, getting them to agree to be the guinea pigs in our grand experiment won’t be easy either way. Maybe they aren’t so excited. We’ll have to tell them about the whole arm-health and hitters-sitting-dead-red part so they buy in. Maybe if we tell them they’ll be way out in front when the robot umps come — that they’ll be the most ready to take advantage of the new parts of the zone that will open up then — maybe then we’ll get ourselves a pitcher ready to completely flip the script on today’s batters.
With a phone full of pictures of pitchers' fingers, strange beers, and his two toddler sons, Eno Sarris can be found at the ballpark or a brewery most days. Read him here, writing about the A's or Giants at The Athletic, or about beer at October. Follow him on Twitter @enosarris if you can handle the sandwiches and inanity.
Clever idea. But if it’s 80% breaking balls then wouldn’t we expect hitters to adapt to that, too? If a pitcher having multiple breaking balls that they could justify throwing 80% of the time that would be great, but the benefit would be that they aren’t throwing any pitch more than 50% of the time and hitters couldn’t sit on anything.
Think having fastballs under 20% would seem more important than breaking balls over 80% regarding getting batters to stop sitting fastballs.
But how many pitchers now have multiple sliders that they can throw? Same with changeups (though they don’t for this exercise). Would it really be so hard to a guy with good breaking stuff to learn a couple different slider and curveball grips? Heck, at the point, a hard slider with mostly vertical break could be the primary offering while a bigger curve with the similar break would then become something of a changeup. Having multiple speeds and different kinds of break seems like it could be enough. And if you’ve got a hard fastball, you can use it to throw off the timing of a hitter who’s sitting 82-88 when you blow it by him at 97-98.
This would be like a cousin of knuckleballers who use that as their primary pitch?
It’s already happened–dude named Greg my senior year of high school. Kid couldn’t hit 80, so he simply threw a curveball (which he could locate quite well for an 18-year old), like, 90% of the time. Totally made 2nd Team All-Conference.
Cool story, bro.
Pitchers often deliberately throw breaking balls out of the strike zone to get behind-in-the-count batters to chase. (Pitchers try it more frequently with breaking balls than they do with fastballs, I guess.)
Probably pitchers could locate their breaking balls within the “wide zone” more than 47.8% of the time if they wanted to.
That 20% fastball would become very important then, it would be there to create the tunneling effect and to keep hitters honest. You can see what happens when a pitcher throws a lot of an offspeed/breaking pitch with what hitters are doing with Shoemaker right now. They are laying off pitches that are low in the zone because they know it is probably his split. I would like for him to throw more fastballs is those counts and mix it up a little bit, make the adjustment to make is other pitches get swung at.
I’m sad to see no mention of Rich Hill in this article. Because he might be the best example, or maybe his curve is just so much of an outlier that you can’t use him. On the health side of things (arm health at least) I don’t know how we didn’t realize this earlier, I mean think of all the pitchers that lasted forever almost all of them didn’t throw a mean fastball (Nolan Ryan being the outlier because he’s a freak of nature)
The thing is though, even at like 50% breaking balls, Hill’s fastball is decent. He throws it infrequently enough that he can freeze hitters with fastballs down the middle. As a guy with a great breaking ball and mediocre fastball, Hill clearly was the perfect candidate to go into uncharted territory with his fastball/breaker ratio, but it seems like his ideal mix is still close to 50-50, and not like 20-80.
Does the chart at the top really show progress? Sure, the text below it describes it as ‘slow and gradual’, but one might also just call it noise. The linear regression line has an R2 of 0.54, but that is with some rounding– comes out to just 0.40 without the rounding.
Keeping the rounding, but dropping 2010 and the R2 drops to 0.11. If including or not including 2010 is the difference between a a trend and just noise, does that suggest anything?
The data goes back to 2007. Actually including all 10+ years gets you a 0.72 R2, so that might have made a stronger case.
But given an R2 of 0.83 from 2007 to 2012 and 0.02 from 2012 to now, maybe it just means there was something going on and it has long since stabilized and this was more of a story five years ago than it should be now?
Stop hurting my brain.
TIL that the plural of eephus is eephi
I can’t statistally back this up but it seems like debilitating blisters have gone up with curveball useage.
Good point, did not think to add that, but did find some sort of link recently: http://www.fangraphs.com/blogs/you-can-probably-blame-rich-hills-blisters-on-his-curveball/
Generally speaking, elbows.
Elbows are the reason we don’t (and can’t?) see that many breaking balls.
The comment below about arm slot also makes sense.
The traditional plural is “Dear pitcher, please throw one eephus. And while you’re at it please throw another one.”
Just read that robo ump article. Is there a rule that you have to have a catcher? If there were robot umps, you could just put 8 players behind the pitcher as long as there aren’t 2 strikes.
I guess you could do it without robot umps, but I doubt the home plate umpire would appreciate it.
there is actually a rule about having a catcher! will try to find later.
MLB Rule 5.02
This reminds me of the various truisms about how as a pitcher, sometimes you have to rely on your “best pitch”, which in those various contexts always refers to the fastball. Clearly, it is conventional wisdom–one which many pitchers and coaches are now questioning–whether a fastball is actually a pitcher’s best pitch. It seems like maybe pitchers really were using their mixes suboptimally, and that there are a whole bunch of pitchers who would benefit from throwing their breaking balls way more.
However, I think there is some truth to that conventional wisdom. To the extent that I have studied pitching, these are the two qualities of a pitch that really show a strong relationship with effectiveness:
1. Pitch velocity (basically the higher the better, but there is some benefit if there is a velocity differential)
2. How differently a pitch moves compared to the other pitches that pitcher throws, or other pitches generally.
If you look at that, only one quality actually helps on its own: velocity. This makes sense, as it directly implies less reaction time for the hitter. Movement is important, but you can see it is mostly important *relative* to other pitches.
This is my point: I don’t think even really good breaking balls are inherently *that* hard to hit. Case in point: a lot of really great sliders barely move at all. So if a pitcher throws like 80% breakers, sure, it may surprise hitters at first, but if we’re talking about a high-70’s or low-80’s pitch, then as soon as hitters adjust to the movement, it will just mean they have that much more time to track the ball. Even Rich Hill, at far less than 80%, throws multiple curveballs from multiple angles to mix it up, and there aren’t many breaking balls in baseball better than Hill’s curve. Could Hill throw even 50% if he only had one breaking ball shape? Could he get away with throwing 80% curves? I highly doubt it would help him, and if he can’t do it, who could?
It’s a good point about velocity, but I’m hoping that fewer HR and injuries is the ‘get’ for ‘giving’ velo. They’d still have to throw the fastball some, and they’d also have to have multiple breaking balls, for sure. But between the cut fastball, the cutter, the baby slider, the Warthen Slider, the regular slider, the power curve, the straight curve, the roundhouse curve… there are a bunch to choose from.
I guess the trouble is how many of each of those can one pitcher throw? If one pitcher could throw them all, that might work, but many pitchers can barely master one breaking ball, let alone two, without them all melding together.
Wasn’t it Mike Fast who found sliders effectively do move a lot, because what matters is motion within the hitter’s field of vision? Not motion relative to the flight path. Sidearm sliders with zero Magnus force are moving laterally by the geometry, while fastballs cancel out the geometry with their deflection and come in straight to the hitter.
I’m sure that is part of it, but sliders tend to have big platoon splits, suggesting what you gain against hitters from one side, you lose against hitters from the other. I’m not sure if the net effect is positive or not.
Either way, that would be a plausible way in which movement is effective in isolation. However, I think that effect is simply small relative to velocity, and relative to the effectiveness of movement as a unique look in a pitcher’s arsenal.
Interesting idea. Doesn’t Perdomo also have an elite breaker? Thoughts on him potentially doing this?
The only problem I have with this is that it’s really easy to lose your arm slot when you over-rely on breaking balls. I know I had issues with it in college, and I was only throwing my curveball maybe 45-50% of the time. Knuckleballers sometimes have similar problems; I know Steven Wright has talked about using his fastball to maintain a good arm slot. This might not apply to some players, but it’s still a possible issue.
Is 80% breaking-balls a joke or just a device to say guys should throw more? I mean if the guys in the graphic are generally throwing a lot of breaking balls at around 50%, another 50%-plus improvement in throwing that many breaking balls seems pretty crazy.
I guess my question is why do you think a given pitcher’s pitch selection is below equilibrium? I’d propose one way to test this is look at guys that changed their pitch selection a lot (let’s say by more than one standard deviation) and how many actually improve their score/rating (use Fangraphs pitch value).
To me, it seems such a change in approach would be tantamount to adding a pitch. I don’t know, it seems like guys fail with a adding a pitch as often as the succeed–and this is already driven by attrition bias to some degree since most guys won’t even try a new pitch unless they are comfortable with its development in the off-season or spring training.
For me it was less about the particular pitchers not being at equilibrium and more about the fact that a hitter could still expect/sit on two fastballs per average plate appearance. I’d like to see what happens if they can’t do that.
OK, so isn’t the test to compare OPS among players who see 2 or more breaking balls (or maybe see more breaking balls than other pitches) in their first plate appearance versus those that don’t? I’d presume you’d want to restrict it to among those PAs against starting pitchers since I can only assume relievers tend to be more reliant on fastballs than starters.
Based purely on my failed high school baseball career, I don’t think it seems like a good idea. The only time I felt confident I’d hit well is if I sat curveball after seeing the first one. Of course, this was high school and taking one good curve ball was the best recipe for seeing a second bad one. That doesn’t happen in the pros.
“how hard they hit the ball in (the)* air”
“Clayton Kershaw probably doesn’t need (to)* change anything up.”
Honestly who edits these posts?
Fixed! Thanks. This was read about five times by two people, poop happens.
Blame it on Carson. Even if he didn’t read it, blame him anyway.
Isn’t this what Rich Hill does every time out? Seemed to turn him from a journeyman to a solid major league pitcher (when healthy). Don’t forget the affect of breakers on blisters though, as Richie Rich knows all too well.
He’s not on this chart, but my first thought when to Hyun-Jin Ryu, who should probably never throw a 4-seamer again.
Sergio Romo is getting close, and it isn’t helping him too much.
Well. A .319 opponent’s BABIP combined with a lowly 56% strand rate don’t help him there.
Interesting article. What about for the Japanese/Korean leagues? I know from reading the Thames articles (and everything else) that guys over there seem a lot more willing to constantly throw breaking balls. Some seem to find success but observationally speaking, you have to have a large quantity of pitches that break differently. The ole DiceK model.
I think you just inspired Terry Mulholland’s comeback attempt.
I think you’re absolutely right that pitchers should use their best pitch as often as possible. Pitchers with bad fastballs and great breakers could see their fastballs actually get better results if they used it infrequently enough. Peak R.A. Dickey got away with a horrible fastball by throwing something else way more often.
At first, I was hesitant to agree with you on the grounds that I thought more breakers == more elbow pops, but if the research shows that’s wrong, then there’s no reason for a pitcher to avoid doing this. Wainwright would be a perfect choice for this sort of thing if he still had his vintage curve and cutter, but he hasn’t thrown either pitch consistently enough to hope for a radical change like this.