The great debate about Matt Cain, his xFIP and his HR/FB ratio, exploded across the internet again yesterday with multiple articles on both sides of the issue with long threads of comments following each article. The spark was a post by Paapfly who looked at Cain's stuff and theorized that his success is based on a rising fastball that induces a lot more popups than the average fastball. xFIP enthusiasts in the sabermetric community seem to a a bit touchy about Matt Cain these days. Dave Cameron posted a study looking at several pitchers who had below average HR/FB 5 years in a row and found that they subsequently regressed close to the mean. Unfortunately, Dave did his math wrong and all but two of his examples actually maintained a HR/FB significantly below league average and the two who didn't saw their careers completely fall apart, probably due to age/injury/loss of skills. Dave later admitted way down in the comments that his math was wrong. He went back and changed the numbers in his original post but failed to acknowledge that the new numbers lead to a much different conclusion. I've already weighed in on the subject in a previous post, but I'll just add two comments here that have clarified in my mind over the last 24 hours:
1. If HR/FB is truly a random occurrence that regresses to a league mean of about 10.5%, regressions should include numbers on both sides of the mean. The pitchers Dave Cameron cited did have a rise in their HR/FB, but continued to maintain a ratio significantly below league mean which strengthens the argument those pitchers, like Matt Cain, possessed a skill which allowed them to suppress HR/FB.
2. Matt Cain's HR/FB has been > 1 SD below the league mean for 5 years in a row. In any single season, there is a 12% chance that any pitcher would fall out of 1 SD to any given side of the mean. The probability of maintaining that variance for 5 consecutive seasons is 0.12 to the 5'th power, or about 1.7/100,000. I'm pretty sure there have not been 100,000 pitchers in the entire history of baseball! Yes, Matt Cain may have been insanely lucky over the last 5 years, but it's much more likely that he possesses a skill at suppressing HR/FB.
Subscribe to:
Post Comments (Atom)
I have been intrigued by the debate as well and there was an idea posted by Dave Pinto at baseball musings that Cain is able to have below average home run rates because of an elite "rising fastball" (one that falls less then expected and in the absence of gravity would rise).
ReplyDeleteI looked further into that and ran a regression that compared a pitchers fastballs vertical movement and velocity to their home run rate and found that there was an association between them.
Essentially a pitcher who throws a hard rising fastball, like Matt Cain is expected to have a better then league average home run rate.
This is enough evidence that I think that it is worth further examination and perhaps enough to take a skeptical look at the idea that HR/FB is purely random.
Completely random side note the verification word when I went to post this comment the first time, dodger. So that is just not right.
Thanks for the input. The rising fastball theory is basically what I've been saying based on my direct observation of a high percentage of Matt Cain's starts. It's nice to see more precise data confirming that. I think that a lot of things we currently attribute to luck or chance are, in reality, very explainable as we learn more about the game. I believe that PitchFx data, once we have a chance to process the enormous amount of information there, will greatly enhance our understanding of the game.
ReplyDelete