Does Fastball Velocity really matter? A look at CC Sabathia’s slowing fastball.

The obvious answer to this question is yes, but I was interested in how much a pitcher’s success was dependent on velocity. Bill Pettit wrote a great piece about CC Sabathia’s fastball, and I hope to shed more light on this interesting question.

(The following pitch f/x number come from Brooks Baseball, and on a technical note I believe that because Brooks Baseball uses a Y_{0} of 55 feet instead of 50 the fastball release speeds are slightly higher than those reported from other pitch/fx sites since they measure speed 5 feet later, so while FanGraphs says CC topped out at 96.3 mph in 2012 Brooks baseball shows him hitting 97.)

I decided to look at all the fastballs CC threw in 2012 to see how well velocity correlated with pitch value. (One more technical note; these are my own pitch value calculations. They differ from FanGraphs in that they treat all balls in play as being worth the same. I did this to try to remove some of the effects of defense. Also note that the scale I’m using is runs- per- pitch and that a negative value is good for the pitcher while a positive value is good for the hitter.) Now according to pitch f/x CC throws a four-seamer and a sinker; because pitch classifications are not perfect I looked at both four-seamers and sinkers together and then looked at them each individually to minimize any artifacts from the classification system. [see the three graphs below]

The first thing that jumps out at me when looking at this is how miniscule the R^{2} values are. Normally any correlation this weak would be ignored but because we know velocity does in fact play a role we can take these R^{2} values to show us how small a role velocity actually plays and how it is just a speck of the big picture. I am not saying that velocity doesn’t matter; I’m just saying that there are a lot of other factors that matter as much or more than velocity. Looking at the plot of all of CC’s fastballs there is no obvious trend (as you would expect with an R^{2} of 0.0002) but the best fit line does have a distinctly negative slope. This means that the faster the pitch the less likely it is to give up runs. A slope of -0.0017 does not seem like much but this is on a per pitch basis so this means that a 1 mph difference would be worth an additional 2.89 runs over CC’s 2012 season (1701 fastballs times .0017 runs per fastball) This may not sound like much but that’s $1.4 million of value so not chump change!

The best fit line for CC’s sinker actually had a positive slope of 0.0056 which would imply that the harder he threw his sinker the worse it did. This seem counter intuitive; a slower pitch is better?, This could be just noise but it does make me believe that Sabathia’s sinker relies less on velocity than his four-seamer does (CC only had 551 pitches classified as sinkers so this is a much smaller sample size so one or two home runs off fast sinkers could throw this off as well). There is a possibility that this is due to a pitch classification artifact but when limiting the regression to only looking at the 100 sinkers with the highest pitch classification confidence the trend actually grows stronger as both the R^{2} and the slope increase.

The four-seam fastball showed the greatest dependence on velocity with the best fit line having a slope of -.0033, which is roughly double that of the slope for all of his fastballs. Another way to look at it, is CC’s loss of four-seamer velocity from 2011 to 2012 cost him about half a win.

Now all these run values have to be taken with several grains of salt because they are based off fit lines with such low R^{2} values and there are many other factors that could be confounding the results, and obviously correlation does not imply causation (also the assumption that the effect of velocity is linear and independent from other pitch characteristics is false but necessary for simplifying this complex problem). It is really impossible to say how much better or worse CC Sabathia would be if his velocity were x instead of y, the values based of the best fit lines are really just educated guesses the only way to find out if CC can continue to be a dominant pitcher with a slower fastball is to wait and see.

All that can be said for sure is that velocity can only explain a fraction of a percent of the variance between good and bad fastballs. This means that there are definitely ways in which a pitcher can adjust to more than make up for the loss of fastball velocity. Now what these adjustments may be is a whole new question. A pitcher can only adjust what they can control, which brings up the question how much control does a pitcher have over the different characteristics of their pitches? Is there something CC can control to make his four-seamer effective at lower velocities? I would say yes based on his ability to throw a sinker that does not rely on velocity for its success.

So does velocity matter? Yes, but so do lots of other things. It’s not clear how much control a pitcher has over these other things that matter; anyone can by mistake throw a pitch on the outside edge of the strike zone, but really, how repeatable is that skill? I’ve never heard of someone throwing 95 by mistake. That’s what nice about velocity, there is no luck involved.