Thursday, July 22, 2010

More on FTP, BABIP, and What They Mean (or don't mean) To Us

As my readers know, I have started using FIP, BABIP, HR%, LD%, and a lot of other non "traditional" numbers when analyzing players. Since I'm using these metrics, I obviously believe that they're useful. However, I have never completely explained where I stand on the advantages - and disadvantages of using these numbers.

Since Toz wrote a fairly comprehensive piece on FIP and xFIP recently, I'll take a look at these numbers and explain how you should - and shouldn't - use them.

FIP (as explained here by Dave Cameron of Fangraphs) attempts to take everything that is within a pitcher's "control" (home runs, walks, hit by pitches, and strikeouts) and apply a factor that is similar to an Earned Run Average. xFIP is similar, but attempts to normalize the HR/FB%, the argument here being that most pitchers don't have control over whether or not a fly ball is going to leave the park or not.

Generally speaking, I think FIP and xFIP are both useful. However, I also think that they're shorthand for a number of metrics that we were already using long before FIP was a household acronym.

The biggest advantage of FIP/xFIP is that it is the shorthand element of it. Rather than sit back and explain that Justin Masterson has been unlucky this year because he's striking out more than six per nine, is allowing less than a HR/9, and has a very high BABIP, I can simply point to his FIP and quickly tell you he's unlucky.

However, the biggest disadvantage to FIP is that some pitchers are historically pitch above or below their FIP. Javier Vazquez is the poster child for this phenomenon. Regardless of what interior metric you're looking at, Vazquez's metrics almost always tell us that he should do better than his actual performance. On the other side, Johan Santana has outperformed his FIP for most of his career.

I use Santana and Vazquez because both are fly ball pitchers, and fly ball pitchers generally give up more HR than neutral or ground ball pitchers. My expectation of fly ball pitchers would be that their FIP would generally be better than their ERA.

Why does Santana consistently outperform his FIP while Vazquez consistently underperforms his?

This is where FIP - and most of the more "advanced" metrics - are currently lacking. We're at the horse-and-buggy stage of this kind of data research. While these numbers are interesting to look at, it isn't enough to say that "Pitcher A's ERA is 0.5 better than his FIP and he's due to regress." Anyone in this industry who generates these numbers needs to do a better job of providing research as to why certain pitchers outperform their FIPs while others don't.

I'm going to continue using FIP and xFIP. I believe that this statistics are useful shorthand and can allow us to make educated guesses as to whether or not a pitcher will get better or worse. But remember that they are just that: guesses. Saying that a pitcher will get better or worse because his FIP tells us so is not a fact and is, in fact, a very dangerous game to play with your season.


Ross said...

Hi Mike. I have been enjoying the blog all season. Thanks for keeping it up. I was struck by how timely this post was. Just a couple days ago, I read Dave's "The Consistently Inconsistent Ricky Nolasco". Dave looks only at Ricky and seems to find that Nolasco consistently underperforms against his FIP and xFIP b/c of his performance out of the stretch. Here is the link:

Obviously, this still begs the question on exactly how to use FIP but it does provide a blue print for how to begin to decipher players who particularly over or under perform their FIP, xFIP, K/BB, etc.

Mike Gianella said...

Boy, I like your blog too. Right up my (our) alley.

Ross said...

Thanks Mike. If you are at all interested in AggPro let me know. We just published another paper on it (improving performance for the rate statistics) that I would be happy to send you.