Friday, August 27, 2010

More Fun and Adventures with ERA and FIP

Two nights ago, Buster Olney tweeted that Trevor Cahill was a strong contender for the Cy Young based on his Won-Loss record, ERA and WHIP. Keith Law almost immediately shot back, pointing out that Cahill is 31st among starters in Wins Above Replacement with a 4.07 FIP. Yesterday, Dave Cameron of Fangraphs wrote an excellent article about non-traditional stats and the Cy Young award.

Cameron's points about how we use (or should use) FIP and BABIP were strong ones. I particularly like his description of FIP as a metric that tells more about what to expect going forward than what a pitcher "should" have done. I must admit that I sometimes fall into the trap of looking at an FIP/ERA differential - in either direction - and saying that a pitcher has been lucky or unlucky. But this is not necessarily so.

As far as awards go, I'll be the first to admit that I look more at non-traditional metrics than I used to but still rely on some of the traditional numbers to gauge whether or not a pitcher is deserving of the hardware. The biggest shift in my thinking is that where years ago I used to believe that an out was an out, today I'll give more credit to a pitcher with a higher strikeout rate for controlling the game, keeping his fielders out of it, and minimizing or eliminating luck altogether.

However, Cahill's 2.43 ERA and 1.00 WHIP shouldn't be dismissed entirely in the awards discussion. While it can fairly be argued that some of Cahill's success is due to luck, it is hard to say that it is all luck and that he should indeed have something more like a 4.00 ERA.

Cameron wrote a piece specifically about Cahill yesterday and put up a quick chart looking at Cahill and Justin Masterson.
Trevor Cahill: 56% GB%, 14.9% LD%, 29.1% FB%, .217 BABIP
Justin Masterson: 62.3% GB%, 14.9% LD%, 22.8% FB%, .344 BABIP
Cameron posted these stats in response to the idea that Cahill's low BABIP was due to his high ground ball rate. Masterson was the opposite case: a pitcher who has an even higher G/F rate but who is unfortunate when it comes to BABIP.

Having watched both Cahill and Masterson pitch on multiple occasions this year, a significant difference between the two is that Cahill has been much better at avoiding the fat part of the plate when I've seen him pitch. Masterson has been far more likely to groove a pitch now and again. When Masterson is on his game, he's incredibly hard to hit...and might have better raw "stuff" than Cahill. But Cahill is far more consistent with the stuff that he has.

That last paragraph is entirely anecdotal. I have no numbers to back it up, only what I've seen with my eyes...which most analysts would reject without hesitation.

Nevertheless, I believe there is something to this idea, even if we cannot yet quantify it. We need to see more data and analysis of pitch quality and pitch consistency before we can simply write off Cahill as nothing but an anomaly. As far as the awards go, the results still should matter more than the expected results, at least in my opinion.

1 comment:

Toz said...

I believe the title of my next post will contain the following disclaimer: "Past performance is not a guarantee of future results."

Essentially, FIP and xFIP are a sabremetric attempt to quantify future results. Working much the same way as a market analyst would, the FIP/xFIP analyst looks at a variety of stats to come up with the long sought after "future performance" indicator.

Forgetting for a moment whether FIP and xFIP are a flawed means of evaluating future performance, they are both certainly misapplied as evaluators of past performance. Using Trevor Cahill as an example, back in June, Trevor Cahill's FIP was likely in the high 4.00s (sorry I do not recall offhand, but this estimate is probably in the range of fairness). The reality is, however, that Trevor Cahill did not put up a high 4.00s ERA over the past two months. Instead, he put up a 3.02 in July and a .92 in August.

As it applies to the Cy Young award, then, do we penalize Trevor Cahill because he "should" have had an ERA close to 5.00, or do we reward him for putting up a year to date ERA of 2.43? Do we leave Cahill out of the conversation because of a "weak" K/9 or an absurdly low BABIP? The answer is self-evident - regardless of what Cahill was "supposed" to do, he has done what he has done, and, frankly, done it better than just about anyone in the American League this year.

As I examine these statistics more and more, I find myself overlooking the flaws in the statistics and instead focusing on the flaws in their application. The off-season months will give us a lot of time to examine these issues in more detail.