Relative Velocities
How many times have you been watching a game or engaged in conversations with fellow fans in which the topic of velocity differential between pitches and its effects comes up? The idea in that tremendous offspeed pitches can enhance a fastball has somewhat grown into the conventional wisdom. It may seem odd at first, but throwing a devastating changeup can make an 89 mph fastball feel like a 93 mph fastball, perhaps even higher.
From a pure logic standpoint it makes sense. Consider this scenario: You are standing on the middle of the road and five cars are driving towards you, one at a time, at say 15 mph, from a great distance. But then the sixth car moves at around 30 mph. You aren’t very likely to know exactly how fast that sixth car is but it is going to seem much faster than its actual speed due to what had previously been experienced.
It was mentioned in the comments section of Dave’s article about Jamie Moyer that perhaps he has been able to strike hitters out or simply continue to get hitters out at his age, with a lack of tangible skills, due to his relative velocity; that is, the difference between the velocity of his fastball and his changeup confuses hitters and deems him a bit more effective than we might give him credit for.
Here are the top ten fastball-changeup dropoffs this year, using the BIS data, and their velocity dropoffs:
Mike Mussina, NYY: 15.8 mph
Vicente Padilla, Tex: 12.3 mph
Zack Greinke, KC: 11.9 mph
Johan Santana, NYM: 11.5 mph
Tim Lincecum, SF: 11.3 mph
Edinson Volquez, Cin: 11.0 mph
Oliver Perez, NYM: 11.0 mph
Javier Vazquez, CHW: 10.9 mph
Barry Zito, SF: 10.6 mph
Tim Hudson, Atl: 10.6 mph
And the bottom five:
Brian Bannister, KC: 4.3 mph
Derek Lowe, LAD: 5.1 mph
Greg Maddux, SD: 5.4 mph
Shaun Marcum, Tor: 5.7 mph
Josh Beckett, Bos: 5.8 mph
Of those in the top ten, the only two that throw below 90 mph, on average, are Mussina and Zito, who clock in around the 84-86 mark. Mussina has been in a relative velocity league of his own this year, throwing an 85.9 mph fastball and a 70.1 mph changeup. Now, pitchers don’t follow a uniform routine of changeup usage: some will throw it much more often than others. Mussina only throws his 6.9% of the time whereas Johan Santana has done so with 26.8% of his pitches.
Looking at pitchers who throw their changeup at least 10% of the time, Johan finds himself atop the relative dropoff leaderboard and new names like Cole Hamels and Brandon Webb join him. The sample of players is likely too small from just one year to find any type of significance or trend within the data, and even then, a regression amongst others would need to be run as a starting point to find what type of effects this dropoff has, which I could perhaps get into in the near-future, but if this relative velocity really matters as much as our logic and commentators suspect, then it should be fairly easy to find a meaningful correlation between dropoff and some form of success.
Print This Post
Rather than being overly wordy, why not just simply state that changing speeds is important, and then give us the list of best/worst?
I’m not sure what you mean. The topic is introduced, then an example is offered in out-of-baseball context for anyone potentially reading this unaware of what the relative velocity refers to, then the small snippet about Moyer ties in how someone like he could potentially be effective. Then the lists come and then there’s a discussion about how, if this really does have an effect on performance, it should be fairly easily identifiable in a study using several years of information… something I’d get into at some other point.
Maybe some words here and there could go, but how about you just read which parts you find pertinent?
What is interesting to note is that if a GM were dealt a rotation of any mix of the top 5 and bottom 5 i don’t think he’d complain.
It would be nice to see the top and bottom 5 or 10 using 10% change-ups as the cut-off.
Ask and ye shall receive! Top 10 with 10%:
Santana, 11.5 mph
Lincecum, 11.3 mph
Volquez, 11.0 mph
Vazquez, 10.9 mph
Zito, 10.6 mph
Hamels, 10.4 mph
Webb, 10.4 mph
Verlander, 10.0 mph
Blackburn, 9.9 mph
Garland, 9.6 mph
And the bottom ten with 10%:
Maddux, 5.4 mph
Marcum, 5.7 mph
Cain, 5.9 mph
Hendrickson, 6.0 mph
Olsen, 6.1 mph
Duke, 6.4 mph
Sanchez, 6.5 mph
Rogers, 6.5 mph
Lannan, 6.6 mph
Myers, 6.8 mph
All told, of the 51 pitchers who qualified with at least 10%, the diff between the 10th worst and 10th best is 2.8 mph of dropoff.
Makes me wonder if a good change-up makes for a good pitcher. Relative velocity may not be the best way to judge a good change-up, but frequency of use would probably better. Couple frequency with use and you have a list of some good arms.
It would be interesting to use a couple of metrics for other pitches, fastball (speed, “rise”), etc.
Scappy, the difficult thing is that all pitches are somewhat tied together, and I’ve had this problem before in determining how to investigate which types of certain pitches are the most effective.
The results are going to be different if he throws 5 straight changeups as opposed to FA, FA, CH, FA, CH, strictly from what the batter is seeing. Consider the car examples, driving towards the person. You’ll have a MUCH tougher time gauging the speed if you see FA, FA, CH, FA, CH, then five straight changeups.
There are also different kinds of changeups. Perhaps Matt Cain is able to succeed because his changeup has significant movement compared to his fastball which overcomes the relatively small velocity dropoff.
I agree that relative velocity is probably not the best way to determine the best changeups, however it is likely one of the key components.
What’s interesting is that Marcum is widely regarded as having a filthy changeup (as evidenced in his results thus far this year), and yet his dropoff is one of the worst.
That could be due to a number of things. Perhaps the movement on his changeup is ridiculous compared to his fastball so it’s not a slower version of the heater but rather a trickier pitch to deduce. Or it could be how he uses it. If I recall correctly, Marcum is one of only 3-4 pitchers to throw 5 different pitches at least 10% of the time.
I’d like to throw out there that deception at least appears to be a big part of a changeup’s success. This was talked about ad nauseam a couple years ago for Santana and how when he throws his changeup it actually looks like he’s throwing a fastball and not (of course) some other pitch.
This is quite similar to serving in tennis. Pete Sampras did not have the fastest serve (thought it was pretty fast), but his serve was so deceptive since his toss was always the same, the motions he went through were always the same, yet you couldn’t tell where it was going or what type of spin he was going to put on the ball.
Difference in changeup vs fastball speed becomes fairly meaningless if you know a changeup is coming and my guess would be the best changeup is a properly disguised changeup.
Exactly, though at the major league level I would tend to think the pitchers are better at masking the pitch moreso than each level of competition prior. So when we’re talking about say a group of 10 pitchers all very adept at masking their changeup, its usage in sequencing and location becomes very important as well.
Ultimately, all of these aspects are likely ahead of the relative speed dropoff, which is why I honestly don’t expect to find any strong correlation between it and success.
Maybe sequencing and location are very important, even with fastballs, and speed doesn’t matter all that much. In that case, baseball “intelligence” and maturity may go at a premium.
I vaguely remember an article over the off-season about a similar topic. It might have been by Joe P. Sheehan or Mike Fast or Josh Kalk or Pizza Cutter or (more likely) someone I can’t think of. Anyway, most of the article broke down individual pitches (by specific pitchers) using PITCHf/x and pbp data, and presented their linear weights. For example, every time Johan Santana through a change-up, it resulted in an average net change of -.03 runs or something like that. Then the author picked out the most effective changeups. Surprisingly, they weren’t on the list of the slowest relative to that pitcher’s fastball. The were about 8-10 MPH slower and all had similar movement. Very interesting stuff. Sorry I don’t have a link.
Sky, sounds like a John Walsh article, maybe his runs/100 metric. From an intuitive standpoint I would expect movement and location relative to the preceding pitch, or even context of the sequence to add to the devastating nature.
I can tell you that despite Marcum’s lack of velocity difference between the change/fastball, his change up just drops out of nowhere. He keeps the same arm angle/speed so it looks like his fastball, but it drops at the last second and also goes down and to the right. So for a lefty it’ll start off in the middle of the plate, and then drop down very quickly to the right.
So not only is it a little bit slower than the fastball, but when the hitters about to swing on it, it’s a lot lower (and farther away/inside depending on whether it’s a lefty/right) than where they expect it to be and they just swing over it.
What also helps Marcum a bit is he throws everything – Fastball, cutter, curve, slider, change up etc. – so the hitter has to be prepared to hit any one of those pitchers. So another thing to look at might be how many pitches they throw, in addition to the % of change ups thrown.
A corollary to this phenomenon is when a fire-balling reliever will come in to spell a soft-tosser, and the announcers will speculate how hard it must be for the batters to adjust. Like maybe David Aardsma relieves Wakefield, or in my neck of the woods this was a common refrain when Robby Nenn (gob bless him) would take over after Kirk Reuter or Mark Gardner or someone would go 8 strong.
Is there any way to look at large velocity differences between consecutive pitchers to see if a similar effect occurs? Like is David Aardsma in fact more effective when he relieves Tim Wakefield? I’m not totally sure how you would go about this.