Astros' Change Ups

"Hitting is Timing. Pitching Is Upsetting Timing." --Warren Spahn

Why is the change up pitch subject to so much sabermetric analysis?  The change up is the pitch that best exemplifies the art of pitching.  Spahn, quoted above, believed that a pitcher had to out-think the batter.  The change up is a tool for disrupting the hitter's timing and "slowing down his bat."  The "artistic" aspect of pitching is the most resistant to saber analysis. Pitch sequencing, pitch selection, changing speeds, reacting to the hitter's approach--these factors are all difficult to measure.  So, it's not surprising that analysts would turn to studying the usage of change ups,with the potential for understanding how a pitcher becomes a pitcher rather than just a thrower.

At Baseball Prospectus, Harry Pavlidis--one of the well known Pitch f/x experts--wrote a three part series about change ups.  I found Part 2 and Part 3 to be the most interesting.  I won't repeat all of his findings--you can read the articles to get the full story.  But these points intrigued me the most:

• Pitchers with high velocity FBs are more likely to get whiffs on the change up.
• A large gap between FB velocity and change up velocity is more likely to produce whiffs on the change up.
• A smaller gap between FB velocity and change up velocity is more likely to produce ground balls, instead of whiffs, off the change up. 
  
• Higher usage rates and change up whiffs are statistically related.
• Some vertical drop is useful in producing more groundballs off the change up.
• The top change up pitchers get the best of both worlds, whiffs and ground ball contact.

With these points in mind, I will provide a brief display of change up usage and outcomes among Astros' starting pitchers this year.  Below are the usage rates and change up velocity for Astros'  starting pitchers (as starters) for the change up.  The pitchers are ordered according to the run prevention value of their change up in 2013.

As noted in the B-Pro article, higher usage of the change up and whiff rates are statistically related.  This could be self-selection: pitchers know their change up is effective at getting swings and misses, so they use it more.  Or it could reflect the possibility that the change up becomes more effective if it is used more.  Perhaps the hitter has to be convinced that there is a significant probability of seeing a change up before it adversely changes his approach to the at bat.  Most likely, it's some combination of both reasons.

We have discussed the effectiveness of Oberholtzer's change up in previous articles; not surprisingly he throws the most change ups.  Lyles and Peacock are the least likely to use their change up.

The table below displays the velocity differential between the pitcher's FB and Change Up, as well as some relevant pitch outcome data for the pitch.  I have also provided the 2013 MLB average, so that you can determine whether the outcomes are above or below average.  Again, the pitchers are listed in order of change up run prevention value.

Clemens has the highest change up pitch value per 100.  But this ranking should be taken with a grain of salt, because the data is limited to his work as a starter, which is both recent and a small sample.  But maybe this shows why the Astros want to look at Clemens as a starting pitcher.  As a reliever, Clemens probably has less opportunity to use his change up effectively---and this is tentative evidence that his change up may be pretty good.

Again, it's not surprising that Oberholtzer has the next highest pitch value for the change up.  He can throw it for strikes, it produces the most whiffs, and it also ends up in play a lot, where it has produced outs.  Oberholtzer seems to fit the "best of both worlds" outcomes with his change up, with above average whiff rates and weak ground ball outs.  Oberholtzer has both above average velocity differential and above average vertical drop (not shown on the table).

Cosart should fit the profile for a pitcher who gets a lot of whiffs off his change up--high velocity FB and a large gap between FB and change up velocity--but he doesn't.  My guess is that he doesn't throw enough of the change ups for strikes (48% strike percentage vs. 61% league average). This also probably reflects inconsistent command of the pitch.  Hitters aren't going to whiff a lot on pitches that they don't expect to end up in the zone.  If Cosart can work on commanding his change up more, I would expect a significant increase in his overall K rate.  Given the low strike percentage, the positive run value for his change up is somewhat suprising---and encouraging about the potential improvement if he can command the pitch better.

Humber seems like a good example of the pitcher who throws the change up harder and uses a small velocity differential in order to get weak ground ball contact.  I would expect the pitch values for high "in play" change ups to be more volatile from year to year, because they share some of the BABIP good or bad luck characteristics.

Both Lyles and Peacock throw hard enough, and produce a significant enough velocity differential, that one would expect more effective results from their change ups.  Given the below average strike percentages, I suspect that both pitchers (like Cosart) have command issues with the pitch, producing low whiff rates.  If the pitcher is inconsistent in commanding the pitch, they are less likely to throw it in 3-2 or 2-2 counts, when it might be more likely to produce whiffs.  Peacock's change up was effective in last night's game, and he had a very good game against the Mariners.  If he can command the change up, that could take his performance to a different level.

Part III of the B-Pro article raises the interesting point that some pitchers should be encouraged to throw harder change ups in order to produce weak ground ball contact---even though this may run counter to traditional instruction to concentrate on slowing down the pitch.  Not all pitchers can produce whiffs off their change up, and perhaps a a smaller velocity difference (like Humber's) is preferable in order to seek weak contact.  The article points out that an intentional effort to "slow down" the pitch may reduce the deception of the pitch, with a resulting arm speed difference that can be identified by the batter.

Referring to a presentation by ML pitcher Brian Bannister, Pavidis says:

Bannister recognized this point both in our conversation and in his presentation: How do you get what seems to run counter to conventional wisdom into the hands of young pitchers? What if an organization can find the "throw your changeup harder" candidates and effectively and consistently train them to do so? It may not be the path to missing bats in the minors, but for some pitchers (like Bannister), missing bats in the minors did not translate to doing so in the majors.

Now, we have a concrete and attackable problem space. Statistical and anecdotal evidence suggests that some pitchers will find more success by abandoning what may be the conventional profile for a changeup. Organizations always want more return on their player investments by turning more guys into effective big leaguers that would otherwise be churned out of the system. You can never have enough pitching.

This is a provocative point of view.  I wonder if the Astros' farm system instruction has considered this idea?

Do you see anything else in the change up data for Astros' pitchers that's worth noting?

More from Crawfish Boxes:

• Brad Peacock Keeps The Mariners In Check As Astros Bats Stay Hot.
• Buying the Astros' Plan Lock, Stock & Barrel
• AMP Mailbag: A new bag of Astros questions
• Jarred Cosart Shut Down
• Houston Astros Minor League Recap (9/10/13): The Legend of Carlos Perdomo