Does this subject go back the boogeyman of the BABIP theories popularly touted which assert that given steady K ratios allowing hits is simply luck? I like Sheets a lot, I'll agree with that. But as you know I don't like the theory I mentioned above.

I'm still waiting for the "Joe Blanton - Baseball's Unluckiest Pitcher." Look at his BABIP last year vs. this year - he's not worse, he's just clearly unlucky. The vaunted Oakland defense must just give up behind him or something.

What are your thoughts on Joe Blanton, am I right, do you think he's simply the victim of bad luck and not at all less effective as a pitcher?

You extolled the virtues of Jake Peavy, but watching him last night vs. the Dodgers, it isn't that he isn't unlucky, he simply isn't the pitcher he was last year.

Some people have theorized that pitchers cannot control home runs as well as you think, that they can control flyballs but not home runs, and that if your flyball/home run percentage is above 10.5% you are simply "unlucky." Tim Hudson's BABIP is still the same, his G/F ratio is still the same but his F/HR percentage has gone from 6.5% in 2004 to 17.1% in 2005, to 16.5% in 2006. So he would appear to be "unlucky" for two years now. But is anyone buying on a return to the good 'ol days?

And even more striking example, Mark Mulder's BABIP this year is firmly in "unlucky" territory, a steep increase from last year. His F/HR percentage has jumped from 15.3% last year to 25.3% this year. According to the theorists, Mark Mulder is simply "unlucky." But I think anyone with eyes knows that isn't the case.

I thought Sheets was taller than 5-7!

In response to APing's comment above, I have never said pitchers have no control over batting average on balls in play. Instead, I believe BABIP is determined jointly by the pitcher, his defense, the ballpark and, yes, a certain level of randomness or luck, if you will.

I never mentioned Joe Blanton, Tim Hudson or Mark Mulder in this article and, unlike Jake Peavy, don't recall extolling their virtues. If anything, using BABIP, Blanton's regression was somewhat predictable. His .255 BABIP in 2005 was simply unsustainable and his 4.61 FIP (rather than his 3.53 ERA) proved to be a better indicator of his future performance.

With respect to Hudson, you are once again pointing out just how effective FIP is in evaluating a pitcher's performance. His FIP is essentially the same from one year (4.51 in 2005) to the next (4.59 in 2006) based on relatively stable K, BB, and HR rates. The only real change has been in his ERA, which helps prove the point that it is less effective than FIP in evaluating and projecting the performance of pitchers. (By the way, the increase in Hudson's HR/FB rate the past two years is partly a function of the fact that he was traded from Oakland to Atlanta prior to the 2005 season as anything else.)

Mulder's ERA last year was also better than his peripherals. As such, he is another poster boy for the virtues of FIP. (As a side note, although I'm not here to defend him, Mulder has experienced arm troubles this year. From 5/28 until he was shut down after his start on 8/29, he allowed 11 HR in 26 IP. He was basically throwing batting practice during this span. His last three starts [7 IP, 25 H, 23 R/ER, 2 SO, 7 BB, and 3 HR] had a huge effect on both his FIP and ERA.)

Perhaps without even realizing it, you are showing just how much more effective FIP (or the focus on K, BB, and HR rates) is than ERA in summarizing how well or poorly a pitcher has performed and is likely to perform in the future.

You don't have to buy into the "strong form" of the DIPS argument to find FIP useful. Even before DIPS came along, it was easy to see that strikeout and walk rates were a better reflection of a pitcher's true skill than ERA in a given year. In fact, that's been one of the more common complaints about DIPS -- that it didn't really uncover anything new.

Pitchers don't get just "lucky" with BABIP. They also get "lucky" with strand rates, bullpen support and probably one or two other things. That means that you need to regress a pitcher's performance to a sample mean, something FIP basically does for you.

I ran a very simple regression of last year's FIP vs. this year's ERA (for pitchers with at least 100 innings both years) and found that last year's FIP was a much better predicter of this year's ERA than just using last year's ERA. I found a variance of .07 for FIP vs. .09 for ERA (a more significant difference than it sounds).

But the more important point is that, even when you regress last year's ERA against this year's, the regression equation calls for you to just take half of last year's ERA and add it to a constant of 2.50. This is simply another way of taking a single year's performance, taking out some of the randomness and regressing it to the mean.

FIP is still a better predicter than the ERA regression formula, but only slightly. This shows what FIP does: it takes out the randomness of other factors and regresses the performance to a sample mean. That's what makes it so useful.

Another way of looking at it is to treat the FIP as a piece of evidence on a question- has Sheets' pitching ability deteriorated as a result of his health issues as his 2006 ERA might suggest? In his case, the FIP provides pretty good evidence that it has not, and that his 3 year ERA is a better indicator of his current ability than his 1 year ERA.

I was making more of a point about BABIP and such, I suppose. Moving to Atlanta can't account for all of Hudson's home run problems, and in 2004 his FIP was still 3.45. Even looking at Mulder's more rudimentary numbers it became apparent something was wrong in 2005. But the "hits are a function of luck" theorists would have to conclude that Mulder was just "unlucky" when he was throwing batting practice.

However, back to FIP, virtually every year Barry Zito defies FIP expectations causing analysts to bang their heads in frustration, it would seem. His 2003 FIP-ERA differential was over 1.20. That's near ridiculous. That's higher than Jarrod Washburn in 2005. His 2004, everyone was all smiles as his FIP caught up to him. His 2005, he beat his FIP by half a run. In 2006, he's beating his FIP by 1.05, which is pretty high. So he's made a career out of it. It's quite likely if Barry Zito stayed in Oakland the rest of his career he'd make the FIP predictions look bad year after year. Looking at Zito's FIP, putting aside a better age and health record, Zito is a worse bet to continue what he's doing than Kevin Millwood last year.

I think Zito can continue to defy FIP *in the right conditions.* If he goes to a team like the Red Sox or Rangers I wouldn't be surprised to see his ERA climb to 5.50.

I also noticed FIP makes Carlos Zambrano out to be fairly mediocre. Apparently every year Zambrano magically defies FIP by a hefty margin. FIP says Greg Maddux was far more effective with the Cubs than Carlos Zambrano. Yeah. That's nonsense. Harang is a far better pitcher than Zambrano according to FIP. Felix Hernandez is also better than Zambrano this year. Now that looks like a substantial crock. You can say every good thing in the book about Hernandez's bright, bright future, but telling me he's a better pitcher than Zambrano this year, no, I don't buy that at all. What are we saying? "Ignore his 4.70 ERA. Ignore his higher home run rate. Ignore his 1.39 WHIP. All of that stuff is just Stone Age smoke and mirrors. Hernandez has a superior BB/K ratio. So he's better. In this new era, a pitcher's job is to wow people with BB/K ratios, not prevent runs or anything."

Dave Bush, Brad Penny, also better than Zambrano. Jake Westbrook is smokin' him!

Javier Vazquez is also better than Carlos Zambrano according to FIP.

This Javier Vazquez business has gone on forever. Ever since 2004, analysts have written dissertations on how Javier Vazquez is really a better pitcher than his numbers. Where is this better pitcher? Is he ever going to show up? How useful is it to continually assure everyone a pitcher is better than their numbers when the better numbers don't come?

Kenny Rogers is another guy who isn't very FIP friendly either.

So a metrics-oriented GM would dump FIP-cheaters like Barry Zito, Carlos Zambrano, and Kenny Rogers and stock up on FIP masters Jake Westbrook, Greg Maddux, and Javier Vazquez. The Cubs, for example. Jim Hendry should forget about overpaying to keep Carlos Zambrano, and trade for Jake Westbrook, who is undervalued and FIP assures us will be better than Zambrano anyway.

If I've laid on the scorn too thickly, I apologize, I've simply encountered a lot of smug bloggers who constantly assure you the counter-intuitive is true with some shady looking theories. I don't think of myself as old-fashioned, when someone tells me righthanded pitcher A:

4.15 ERA, 1.42 WHIP, .295 BAA, 5.08 K/9

is in actuality better than righthanded pitcher B:

3.38 ERA, 1.31 WHIP, .210 BAA, 8.73 K/9

Than I get skeptical. Jake Westbrook is a nice pitcher, but he's not Brandon Webb.

If I come off as exasperated, it's because of my experiences on other boards/blogs. Javier Vazquez has what can only be described as a die-hard cult following of sabermetricians who are hell-bent upon proving, no matter what numbers he puts up, that he's actually a great pitcher. I'll admit he has what it takes to turn it around but I'll dread the day he does it, he's had 3 years of "unluckiness."

Well, you're certainly right that FIP is no magic elixir. It's just one piece of evidence. I'd say that one year of FIP is better than one year of ERA, if you know nothing else about the pitcher. But it could also be that three years of ERA is better than three years of FIP. And, as Mike said, FIP may help you find changes in a pitcher before his ERA changes.

That's the key. A big difference between FIP and ERA should encourage you to look deeper into a pitcher's stats or scouting notes.

Zito, for instance, has a much higher LOB% this year than in past years. I wouldn't look for him to maintain that pace. Last year, he had a favorable DER behind him, unlike the previous year or this year.

Those are the real reasons I'd be pessimistic about him next year, not just because FIP says so.

Clearly, Sheets will not continue to post a .349 BABIP. So yes, he's better than he's looked this year, and FIPS helps tell us that. But Sheets has a career BABIP of .310, and has never been below .290 in a season. Peavy (.298) and Javier Vazquez (.308) also aren't very impressive. In contrast, you have a guy like Zambrano at .277, and Santana and Carpenter have been similarly effective in their current incarnations (past three years). Yes, some of this is team defense, but a significant part is also pitcher skill. Sheets is at best average on this skill, probably worse, while some other pitchers are clearly better. And it matters: the difference between Sheets' .310 and Zambrano's .277 is about .75 in ERA.