Discussion: FIP vs ERA

[uva3021]

There was an interersting article, at Fangraphs, on the comparison between FIP and ERA, stemming from an assessment of Trevor Cahill's Cy Young prospects. He's among the league leaders in WHIP and ERA, is on pace for 18-20 wins, and has pitched extremely well lately, all have shaped his case as a legit Cy Young contender.

But his FIP is ranked 31st in the AL, 4.07, and that ERA to FIP discrepancy is owing in large part to his absurdly low .217 BABIP. Though the argument is, which is a better indicator of prior performance, FIP or one's actual ERA? There have been plenty of studies to suggest a pitcher's current FIP is a more precise indicator of future performance.

And if you are afforded the luxury of having access to a retrosheet database, just run a regression of the two variables onto year end ERA, and the results are overly supportive of FIP being a better predictor of future ERA.

Here is the article

FanGraphs Stats and the Cy Young Award

...But FIP was not designed to give us a better insight into what actually happened, but, instead, what is likely to happen in the future. FIP is part of the collection of metrics that do a good job of predicting what will happen in the future by focusing on things that are under a player’s control. FIP was never designed to be a backward-looking metric designed to tell us what actually did happen. And there’s a decent argument to be made that the Cy Young award should be awarded based on what did happen, not on what should have happened or what will happen in the future.


In Cahill’s case, the real sticking point is his BABIP, which currently stands at an absurdly low .217. The next lowest batting average on balls in play for an AL starter is C.J. Wilson, at .263. The gap between Cahill and the rest of the league is enormous, and it is the driving force behind his low ERA. That .217 BABIP is not sustainable in any way, shape, or form. Unlike things like strikeout rate, BABIP is simply not a skill that a pitcher has much control over, which is why it’s not included in the FIP calculation.


But it is highly unlikely that Cahill has had absolutely nothing to do with his low BABIP to date. Yes, that number is driven down by a combination of outside factors, including his home park, his defense, and some bad hitting by his opponents, but it would be folly to assume that Cahill hasn’t had anything to do with hitters having a tough time getting hits off of him so far. We certainly should not give him credit for all of the hit prevention, and we should not expect it to continue, but logically, I think we have to assume that he has contributed, at least in some way, to the amount of balls that have found their way into the gloves of his defenders. Perhaps he has just hit his spots really well for several months – history says he can’t keep doing it, but do we want to assume that he’s had nothing to do with the results? I don’t.
...

The article itself contradicts Fangraphs' implementation of WAR. Claiming FIP is a poor barometer of past performance is counter intuitive to the very calculation of Wins Above Replacement. FIP is the pitching metric for which all other variables coalesce to when conducting a starting pitcher's WAR

[IrishTim]

Good question. I really only find statistics valuable is they can be used in a predictive fashion. For example, there isn't really such a thing as "hot" and "cold" hitters. If you randomly pick a hitter in the middle of a "cold streak" how would you expect his OBA to look in the next at bat? If you said his average OBA (adjusted for opposing pitcher), you'd be correct. The issue of "clutch" is a bit more hazy, but I generally believe that if I am a manager and it's the bottom of the 9th with 2 outs, bases loaded, down by 1 - I want my best hitter up, not my most "clutch".

Anyway, what this means is that "hot and cold" or "clutch" are more than likely just statistical fluctuations - also known as luck. I know the Cy Young is supposed to reward the pitcher who performed the best over the course of the season. But ERA and W-L record figures luck into the equation a lot more than FIP does and I don't think the Cy Young should be given to the "luckiest pitcher", but instead the best.