Does the Changeup Have a Strikeout Problem?
There is one pitcher out there that throws his changeup over 30% of the time and calls it a ‘heavy sinker.’ Alex Cobb aside, though, we traditionally lump the changeup in with the slider, the curve, the splitter — it’s not a fastball.
And yet, in some really important ways that go beyond movement and leak into usage, the change works like a sinker. In a league where strikeouts rule, the change actually has a strikeout problem.
We know that by movement, the change is most like the sinker. The average sinker has two more inches of horizontal movement, but the league’s changeups and sinkers have identical vertical movement. We know all about this — the beauty of the change is the difference in velocity.
Here’s the problem. In certain counts, the batter expects to see junk. Once the pitcher has two strikes on him, he goes into protect mode. Foul off the hard stuff, but look for a pitch in the 80s with movement, because that’s what the pitcher is going to throw. The percentage of fastballs a hitter sees with no strikes is 60%. The percentage of fastballs a hitter sees with two strikes is 50%.
Perhaps that anticipation with two strikes is why a great changeup alone is not enough to make for a great strikeout rate. Look at the relative correlations between a pitcher’s swinging strike rate on their changeup and their overall strikeout rates.
| Pitch | r with K% | r^2 with K% |
|---|---|---|
| Slider swSTR | 0.398 | 0.158 |
| Curve swSTR | 0.381 | 0.145 |
| Change swSTR | 0.301 | 0.090 |
Minimum 100 changeups thrown, showing r and r^2 between pitch type swinging strike rate and overall strikeout rate in 2014, n ranging from 207 to 291 per pitch type.
So, while your slider’s swinging strike rate explains about 16% of the variance in your overall strikeout rate, the changeup comes in short of 10%. Quantity thrown has something to do with this, of course. The slider is thrown 13.7% of the time, the change 10.4% of the time. But since we focused only on pitchers that throw the pitch, some of that is mitigated (there are more slider throwers as well as sliders thrown). And the difference in correlations is larger than the difference in quantity thrown either way.
Is the change not a great strikeout pitch? Let’s look at the swinging strike rates on these three pitch types in two-strike counts compared to the rest of the time.
| Pitch | 2 Strikes swSTR% | Non 2 Strikes swSTR% |
|---|---|---|
| Curve | 19.1% | 11.6% |
| Slider | 15.8% | 9.8% |
| Change | 9.3% | 6.4% |
The changeup is the worst strikeout pitch among the three standard pitches, and it’s comparatively worse in those situations. In other words, while the curve and slider are both more than 60% better in two-strike situations, the changeup is only 53% better once the batter is a strike away from sitting down.
And yet, there’s virtually no difference in changeup usage by count. The change is used 10.1% of the time in one- or zero-strike counts. It’s used 9.99% of the time with two strikes. So there is still a reason to use a change, even when the batter is more prepared for the difference in speed, and relatively more likely to make contact on the pitch when he’s looking for it.
That reason is probably to get a ground ball. Here are the ground-ball rates for the main three pitch types in two-strike and non-two-strike situations.
| Pitch | 2 Strikes GB% | 1, 0 Strikes GB% |
|---|---|---|
| Change | 70.7% | 49.8% |
| Curve | 55.6% | 49.4% |
| Slider | 49.0% | 45.8% |
Here, the results from the swinging strike inquiry are reversed. The changeup not only has the best ground-ball rate with two strikes, but it also enjoys the largest relative boost compared to other situations. The change gets 42% more grounders with two strikes than it does in other situations, while the curve and slider get boosts closer to 10%.
Ground balls are great, but they are only outs about 75% of the time. Strikeouts are outs 99% of the time. Perhaps, instead of asking slider pitchers about their change as they come up in the system, we should be asking changeup pitchers about their breaker.
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.
Maybe the strikeout has a changeup problem.
Maybe the changeout has a strikeup problem.
no the change up has a change up problem. I love hitting change ups, they are sucky fastballs, with less movement than a two seamer.
I love it when bigshot athletes post on Fangraphs.
jarrod parker.
See: Henderson Alvarez.
Henderson Alvarez is awesome. His Sinker,Changeup, and Four-Seamer are identical to that of Felix Hernandez in velocity,movement, and release point. He doesn’t have breaking ball and can’t strike people out.
Chris Capuano is another example.
Actually no, Alvarez’s pitches are not identical to Felix Hernandez’s. Felix’s change up had a SwStr% of 17.6% in 2014. Alvarez’s produced 9.6%. Felix’s averaged 89mph and Alvarez’s averaged 87mph. Also, the vertical movement on those pitches are completely different. It took 30 seconds to see that you completely mischaracterized just one of the 3 pitches you mention. I wonder what would happen if I looked at the other 2 pitch types …
You are using the Fangraph’s pitch f/x data which is wrong, trust me. If you look at Brooks(which is corrected data), it is very very similar.
PItcher Velocity(mph) H-Movement(in) V-Movement GB% Whiffs/Swing%
Hernandez 90.02 -6.33 1.43 73% 42%
Alvarez 89.86 -6.56 2.10 65% 27%
Grounders may be outs only 75% of the time, but occasionally they’re 2 outs.
So are flyballs and line drives, so what exactly is your point?
In which we learn that Nathaniel Dawson does not take kindly to analyses of double plays that only cover around 85% of all scenarios.
Yes but a ground ball is a double play some percentage of the time and a strikeout is not.
I mean, I am with you on your premise, there is also some game theory in here, but it’s not entirely irrational to throw a changeup rather than a slider/curve when you’re at two strikes.
Also might be a factor that the changeup is easier to throw for a strike on a full count? Not sure of the strike% on each of those but I think the curve is lowest.
A strikeout is a double play some percentage of the time.
yes, a an extremely low percentage of the time
“it’s not entirely irrational to throw a changeup rather than a slider/curve when you’re at two strikes.”
As MGL would point out, you have to throw changeups some portion of the time. The pitches only have the swing and miss rates they do because pitchers presumably are throwing them with a frequency distribution approximately at the optimal strategy distribution.
Two questions:
Can the data be looked at in terms of ‘turnover’ of the lineup. Is a changeup’s effectiveness at all a function of how many times a pitcher has faced the individuals in a lineup during single game.
Can you infer that, via the GB, the changeup keeps pitch count down and increases the number of innings a pitcher’s FB is effective.
Would the changeup’s effectiveness change as the lineup turns over in a way that is disproportionate to how the effectiveness of the other pitches changes? Wouldn’t all pitches be affected?
I’m also not sure to what extent the increased GB rate is offset by the lower swinging strike rate when talking about pitch count. There is a myth that getting a lot of strikeouts makes a pitcher more inefficient. However, strikeouts actually reduce pitch count by reducing the baserunners allowed, because nothing really drives up a pitch count like not getting outs.
I think the idea that strikeouts lead to higher pitch counts might be attributable to a false causality. High-K pitchers are more likely to also walk a lot of batters, and walks have a huge effect on pitch counts. If you look at the P/IP leaderboard on mlb.com, you’ll see virtually no correlation between pitches per inning and K/9. In fact, the same pitcher lead baseball in highest K/9 and lowest P/IP. However, you will notice a huge correlation between P/IP and BB/9.
And yet when you see a batter flail helplessly for strike three against a well-thrown change up, it is a thing of beauty. Maybe its rarity makes you appreciate it more…
/remembers fondly the Marcum change up because he had not much else
Johan Santana and Brad Radke were both awesome to watch when they were with the Twins. Santana was great because he rang people up on it so easily and Radke because you occasionally got to watch batters swing and miss on an 85 mph fastball because they were expecting a changeup.
Trevor Hoffman and Keith Foulke were fun, too.
And the once-seemingly-ageless Jamie Moyer and his 80-82 mph fastball that would lock hitters up on the inside corner.
Francisco Rodriguez is an interesting case. His change-up drives almost all of his swinging strikes (and consequently, strikeout rate), as none of his other pitches is even average by swinging strikes and his changeup is amazing (29.5% swStr). Also, his changeup only got 43% groundballs last year, so it’s not getting grounders.
The things that are lost in almost all these discussions about pitch are reasons why kids are usually encouraged to learn a changeup before they learn a breaking ball: arm strain, pitch count, and opposite-handed batter advantage.
Pitch count should be pretty obvious: if you can get an out in 1 or 2 pitches, why would you go for a strikeout in 3+ pitches? Sure, there’s a level of risk in allowing contact, but saving pitch count surely has some value. Maybe more in Little League with the new rules, but still some value regardless.
Arm strain is less obvious in the modern age but something that was a huge topic when I was growing up. No pitch, thrown with “correct” (however the hell you define that) mechanics, should tax the elbow, shoulder, forearm, etc. more than any other pitch. However, it is easier to throw a curveball or slider with “incorrect” mechanics, since the release motion is inherently different than that of the fastball, which the changeup is not (usually). The slider and cutter in particular can naturally lead to increase forearm strain, leading to extra strain on the tendons/nerve/ligaments on the inside of the elbow.
Lastly, we have the thing that everyone brings up in articles about starting pitchers: changeups provide the necessary weapon against opposite-handed batters. Pitches that move away from the batter are generally more advantageous than those that move towards the batter. Opposite-handed batters have a natural advantage already against pitchers, simply because pitches a moving towards them and have a more visible release point. The changeup is the pitchers attempt to neutralize that advantage. Since most changeups (I would venture) are thrown to opposite-handed batters, it makes sense that K rates are down for changeups, because the pitcher is already working with a natural disadvantage.
I agree with this, but this seems really easily testable, I wonder if anyone has tested it. If change-ups offset the advantage of opposite-handed hitters, then pitchers who throw more changups should have less pronounced platoon splits. Does anyone know if that’s the case? And if its not the case, is there something wrong with my methodological proposal for testing that which I’m not grasping at the moment?
I think it has been observed. Ricky Romero is the guy that comes to mind for me who threw mostly changeups and curveballs, and he had a huge reverse split.
The risk is that you won’t retire the batter you’re pitching to, and if you don’t retire the batter, then you don’t actually save pitches.
The extra pitches it takes to get the K on the whole pay for themselves by the fact that none of the batters you K will reach base and extend the inning.
The thing that really drives up pitch counts is throwing balls. If a pitcher doesn’t have strikeout stuff and tries to strike everyone out by throwing junk whenever he gets two strikes, that will not be efficient. However, really great strikeout pitchers who have the ability to go in and out of the zone will be as efficient as any GB pitcher.
The article also didn’t discuss what happened on the balls in play, HR rate, etc…
I’ve noticed that anytime a pitcher throws a “bp fastball” to get a roll over gb, it’s tracked as a change up……..heck, any badly thrown breaking ball that does not move is tracked as a change up.
That’s true and might be the reason for the low swSTR on ‘all’ changeups (we found the average regularly-thrown change had a 13% swSTR), but with two strikes… I doubt those are a lot of BP fastballs. And the relative increases also point to something else going on.
ok, i’m not a numbers expert – but throwing the bp fb with 2 strikes is one of the better times to throw them.
I found there are a lot of classification errors outside of Brooks on changeups. For example, I was writing on Henderson Alvarez and tried to gets a Baseball Savant heatmap on his change up and it kept giving me pitches that were in the 85-88mph range, which isn’t right since he averages 90-91 on his real changeup.
If that’s true, then what is the pitch that he’s thrown hundreds of times that hovers around 86 mph and has very little rise and lots of armside movement (typical of a changeup)?
It is hard to say Bip.
But Brooks clearly has a the Change averaging 89-91 on a game by game basis with considerably less vertical break. My guess is that the the Sinker,Fastball, and Change blend together in the velocity. If one is thrown poorly, it gives the wrong reading
I think what we can learn here is that you should just have a dominant slider and changeup instead of just 1 of the pitches.
But yes, good points made in the comments. The opposite handed effectiveness is huge. Carlos Martinez with the Cardinals talked with Pedro about his changeup, and with Carlos’ problems with lefties, it would be pretty amazing if he learned a good changeup. Also, I’d be interested in seeing a study that shows if changeups can make the rest of a pitcher’s arsenal more effective (such as the Radke example given earlier).
The handedness issue also brings up another point that sliders are often used exclusively on same-handed batters, where-as a changeup can be fairly effective against either side. If a pitcher has a slider and a changeup in their arsenal, and they just can’t get a “feel” for their slider that day, their changeup can help them to at least eat some innings.
I think another good point about changeups is that guys with hands that are thicker and not as long of fingers usually can get a really good feel for changeups, but with breaking pitches they often can’t “get on top” of their pitches and end up spinning meatballs to the batter. The changeup is an important pitch. As stated, it’s a safer pitch for younger guys to learn, and it also helps pitchers become more comfortable with changing grips so that down the road they can learn breaking pitches more easily.
Every good pitch should make a pitcher’s arsenal more effective compared to that same arsenal without it. Even if the pitcher adds a curveball that is nothing like that pitcher’s fastball, the fact alone that the pitcher has another thing to go to means he can save the fastball for times when it is more effective. So the question is, does a changeup improve the rest of a pitcher’s repertoire more than the addition of another kind of pitch would?
Also, I’m no pitcher, but I have tried to throw different pitches before. I have very small hands, but personally, I find it much easier to throw a curveball than a changeup. I can actually get on top of the curve and at least get more or less the correct spin axis (velocity and spin rate is another story)
Now try throwing that curveball 75+ mph with decent deception in your delivery. It’s too easy to let your hand drift to the side and not get the right spin. Guys with longer lankier hands tend to control spin much easier and struggle with keeping a consistent grip on their changeup to really command the pitch.
I caught competitively for a long time and found this to be the case even with the few major leaguers I got to catch. The guys with thick shuvles for hands seemed to command their changeup much better since they could keep the same throwing motion but just change their grip and move the ball deeper into their palm.
This is just a generalization though. I knew some smaller pitchers with smaller hands that could snap off sliders/curves all game long with good velocity, break and location.
Boy if i only could
I’m always a little bit skeptical of pitch-type analyses, because I think its really hard to capture contextual usage. To me the changeup is a very effective set-up pitch to screw with the batter so that you can get him out with something else, for me that was usually fastball. If you’ve put changeup in a batters mind, its a lot easier to get him out with an inside fastball or to generate a chase on high heat, so I always used changeups as a “this is to make the next pitch more effective” pitch, and its of course pretty low risk, if they do hit it instead of the intended swinging strike or foul ball, its very likely to be a grounder.
Long story short, I consistently wonder whether we’re capturing the highly nuanced sequencing that goes into pitching with these, otherwise pitchers would just throw their best pitch all the time. With the changeup, I feel as though throwing a changeup early in a count sets you up to confuse the batter more effectively with something else later in the count.
Interesting, Eno. One thing to consider: Sliders are — I’m guessing — thrown much more often to same-side batters. Changeups are probably better at neutralizing opposite-side hitters. And a lot relievers are fastball/slider guys, which maybe enters into the equation.
As somebody pointed out earlier in the comments, could this at all be due to a platoon advantage bias?
I’ve never viewed the changeup as a strikeout pitch.
Its a timing disruptor.
If you want a strikeout pitch, make it be fast with movement, or just fast, or just a lot of movement.
What would be interested to me would be batted ball velocities by pitch type and player. Foul ball percents on changeups. Results of fastballs following changeups.
One of the big decisions in pitching sequence in my opinion is what to throw after a changeup. If you always throw a fastball, then throwing a changeup will just tell the batter that a fastball is next. And so on.
Rather than look solely at individual pitches, one needs to look at sequencing.
When I threw a change, I was looking to get a topper to the left side or a big time fly yanked way foul. I wanted them to have to hesitate on a fastball or throw something even slower like the deuce.
Anyway, rambling aside. I wouldn’t expect the change to have a high wiff eate or swing and miss % … but I would look for an increase in weak contact or foul ball rate. In my opinion, you’re looking to get the batter to either hit it off the end of the bat, one handed, or have their balance on their front foot where they negate the full power of the lower body.
I think what’s weird about the pitch though is that the problem isn’t that the change doesn’t get swinging strikes, it’s that it gets swinging strikes, but results in fewer strikeouts than the swinging strike rate would predict, compared to other pitches.
I also don’t know to what extent generating weak contact and generated whiffs are separate qualities. If a pitch is good enough to create a lot of weak contact, I would think that pitch could just as easily get a lot of whiffs. It might have to do with location. A really good pitch that a pitcher throws mainly near the zone might generate weak contact but few whiffs, while that same pitch thrown primarily outside the zone results in more whiffs.
Just a thought.
Changeups are also easier on the arm than the curve and especially slider.
If more pitchers were capable of throwing hard sliders more often and have their arms hold up, they would. Hard sliders are just that damn effective.
But most pitchers cannot continually get batters out by doing the same thing every time (sliders on all 2-strike counts).
Joaquin Benoit says Hi