Friday, July 23, 2010

How to use (or not use) FIP

Ross liked yesterday's post on FIP and xFIP, but asked the obvious question:
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.
I agree with Ross, although this was one of the things I was trying to get at yesterday. We can look at the numbers and draw conclusions, but we would need a comprehensive resource (something like Baseball Reference, only for these advanced metrics) going back through at least 20-30 years of data to do the kind of analysis to prove how useful or worthless FIP and xFIP are.

Since I do use FIP/xFIP and I do look at these numbers, perhaps as a starting point I can share with you how I use these numbers.

1) If the pitcher is a veteran, does he historically pitch above or below his FIP?
This one almost seems too obvious. Yet many analysts seldom if ever bother to ask this question. If a pitcher is throwing up an FIP that's half a run higher than his ERA in 2010 but this is close to his career norms, it's not likely he's going to implode. If, on the other hand, he is a pitcher that normally has an ERA/FIP that are identical, I'd worry about regression.

2) I still like LIMA.
FIP is useful, but I still look at the LIMA indicators (K/9, BB/9, HR/9) when trying to decide whether or not a pitcher is going to get better, maintain, or get worse. The K/9 matter most to me, followed by BB/9, and then HR/9. I'd rather have a pitcher who strikes out a batter per inning than one who strikes out six per nine; the pitcher with more strikeouts has more control over the game. The pitcher who walks fewer batters is less susceptible to bad luck on balls in play than a pitcher who walks the park, and obviously a pitcher who allows fewer home runs isn't going to have a high ERA. This leads right to my next criterion.

3) I prefer ground ball pitchers.
While a ground ball pitcher can still be prone to the big inning, a ball hit on the ground isn't going over the fence. The data on HR/FB% and how much control a pitcher has (or doesn't have) over fly balls is murky at best. But I know that a pitcher who has a high G/F rate is less likely to destroy my ERA/WHIP than a pitcher with a low one. Yes, there are exceptions to this rule (Justin Masterson, I'm looking right at you).

4) Is a pitcher "clutch"?
This criterion is at the bottom of my list because it's easy to get sucked into small sample sizes. However, for a veteran who has been around for 4-5 years, we can look at his numbers with men on base vs. with the bases empty and draw certain conclusions. It might just be perception, but certain pitchers seem to perform better with no one on base while others seem to dig deep and pitch well with men on and the game on the line.

As I said above (and yesterday), there is a lot of work that needs to be done in this area before we come to too many meaningful conclusions. However, I recommend all of these points above as good jumping off points when trying to decide whether or not a pitcher is worth buying at auction or trading for in the off-season or mid-year.


Gypsy Soul said...

Hey Mike, how are you? Hope all is well. Re: your first point, I am just wondering how significant a 0.5 differential is between FIP and xFIP when considering career norms, especially for someone who has logged lots of innings. I guess it well might be but I am just not sure and just not a statistician. Your work in this area is great but this isnt anything really new for you and Toz. Be well, Carter

Derek said...

Hey Mike,
A couple points/questions:

1) FIP and LIMA look at exactly the same things (K, BB, HR), but FIP weights them properly in relation to each other. Why bother making the distinction? And why prefer LIMA? Sure, look at the components separately too to get a little closer to the "why," but there is very little differentiating the two.

2) I think it's pretty bold to say that "the data on HR/FB% and how much control a pitcher has (or doesn't have) over fly balls is murky at best."

HR/FB takes roughly 800 FBs to stabilize - that is, to predict just half of the future variation in the stat. That's four full seasons for the average pitcher (and longer for GB pitchers). For comparison's sake, it takes K/AB about 150 ABs to stabilize, or roughly 1/5 of a season for the average pitcher.

And if presenting the data that way isn't your cup of tea, check out this article I wrote last year: To quote myself:

"I looked at all pitchers with at least 12 games started in adjacent seasons from 2004 to 2008. Over this period, we find 63 pitcher seasons where a pitcher's HR/FB strays at least four percent from league average in Year 1. In Year 2, just 5 of those 63 pitchers (7.9%) failed to regress in the direction of league average. That's a very small number, especially when you consider that Chien-Ming Wang (who may be one of the rare exceptions I mentioned) and Brett Myers (who almost certainly is one of those rare exceptions) accounted for 2 of those 5 seasons. Exclude them, and the percentage becomes 4.8%." That's pretty telling, I think.

Sorry for rambling. Discussions on FIP, xFIP, and the like really get me going :)

Derek Carty
THT Fantasy