An Object at Rest

David Murphy has written a very thought-provoking piece for his High Cheese blog for the Philadelphia Daily News. If you follow both the Phillies and Sabermetrics, then you’re probably well aware of the regression lefty J.A. Happ is likely to face in 2010. Those in the non-stats crowd were impressed by his poise last season and think he’ll perform better than do those of us who fancy numbers. Murphy tries to meet all of us in the middle, citing “The Pendulum Effect”:

[…] refers to the tendency of folks on opposite sides of an argument to make increasingly extreme statements […]

Staying on the science theme, I’d like to direct Murphy towards Newton’s first law of motion:

[…] an object at rest will stay at rest unless acted upon by an external force.

All right, so I’m stretching a bit. But the point is sometimes moderates become so concerned with being moderate that they become just as extremist as those on opposing ends. With all due respect to Murphy, I think he is an object at rest and he is willingly guarding himself against being affected by external forces, or the thought processes of those on either end of the stats/scouts spectrum. It’s evident about halfway through when he posts his “moderate’s creed”.

I am not going to go into the math as to why Happ will regress in 2010. Mostly because I’ve researched and written an article about it that may be printed elsewhere and I don’t want to blow my stack, and also because you can probably read the various reasons with some simple Google searching. So if you are interested in the methodology behind that, I suggest staying tuned for my article (I will post details as they come in) or checking out the work of some other writers who have covered the subject.

Simply put, the reasons behind Happ’s likely regression are statistically sound. The Sabermetric stats that frown upon him are based off of DIPS (defense-independent pitching statistics) initially developed by Voros McCracken and DIPS has evolved many times through the years. Each new advancement shines a little more light than its predecessor.

The reasons behind saying “Happ will regress” are facts. This is not akin to a liberal saying that a woman should have the right to choose or a conservative pushing for his right to carry a firearm. Those arguments are based on morals which are neither right nor wrong. Meanwhile, taking one particular side in the Happ debate does not mean you are an extremist; it means you are either right or wrong. Compromising between the two, as Murphy does, still leads to a wrong answer, albeit to a slightly lesser extent.

I don’t want to sound like the brainwashed stat-nerd that mainstream writers so often characterize. However, Sabermetrics have gone a long way towards enlightening us as to the truths in baseball. Many smart people have put in long hours of data analysis to reach these conclusions. As it pertains to the Happ debate, the stats crowd is right: he is likely to regress.

As Murphy points out, it doesn’t mean it’s impossible that Happ will have as productive a 2010 season as in ’09, it just means it’s unlikely. To say otherwise for any reason whether it’s sincere disbelief or to simply attempt a compromise is to be wrong.

Leave a Reply

*

7 comments

  1. Matt P.

    January 25, 2010 02:35 PM

    I’m assuming there will be some regression, but the degree to which he remains an effective starter could surprise people from both camps, and either way it will be a major storyline in the upcoming season. If he comes close to his 2009 numbers, the rotation should saw through the NL.

    Perhaps you’ve covered elsewhere, but the other guy we’ll all be watching in the same regard is Cole Hamels. Does he revert back to his standard form, with many of his ’09 stats showing that he wasn’t as bad as he seemed? I’m assuming so, and obviously that’ll be a monster storyline in 2010 as well.

  2. Aaron H

    January 25, 2010 03:51 PM

    That’s a good point about Cole, but I’m wondering what form you’re expecting him to revert to. If you’re thinking about his ’08 season, where he posted a 3.09 ERA and led the league in WHIP, I think that’s overly optimistic. If you’re looking at something in between ’08 and ’09, like 3.60-3.70, I think that’d be more accurate.

    In general, you’re right though. A full season of Cy Young quality pitching (as oppose to two months of Cliff Lee) with our 2-4 starters, even with their realistic projections, should still be one of the deepest staffs in the NL.

  3. Bill Baer

    January 25, 2010 06:33 PM

    The term “regression” gets thrown around much too liberally, especially by us stat-heads. So I’d like to clear up exactly what is meant by that word.

    Happ’s peripherals pointed to outstanding luck. I’m being intentionally vague for reasons stated above, but Happ was very, very lucky. Most of his key peripherals were outside of one standard deviation away from the mean. We would expect Happ to be within -1 and +1 standard deviation about 68% of the time.

    32% doesn’t sound too bad, right? The odds of Happ enjoying another sub-3 ERA season are very, very low because he would need his peripherals to pan out similarly. So you take X, Y, and Z peripherals and see what the odds of them being outside of +/- 1 SD would be. This is very crude (please don’t copy this — this is not at all close to proper regression, but merely a quick way to make a point), but it’d be .32*.32*.32 and so on for every peripheral Happ will be dependent on. For three peripherals, it comes out to about 3%.

    Additionally, when stat-heads say Happ will regress, it’s not a specific statement akin to “Happ will suck”. It’s a generalized statement that means that Happ will be much closer to his or the league averages than the extremes he showed last season.

    I have forecasts of Cole Hamels and J.A. Happ written, but I’m waiting to see if they will be used elsewhere. That’s why I’m being so ambiguous and for that I apologize.

  4. doc

    January 25, 2010 10:56 PM

    Regarding Happ, if everything stays the same, then yes, he almost has to see his ERA go up. But what gets left out of the discussion is that maybe he’ll find a way to adjust and stay ahead of the curve. No, not likely, but Happ does have some control over this. By the way, I wouldn’t be at all surprised to see Happy and Hamels almost trade ERA’s.

  5. Dan

    January 26, 2010 04:09 PM

    Not to steal anyone’s thunder, but it’s unfortunately easy to show how Happ will regress. He was incredibly lucky with men on base in 2009, which lead to his performance:

    With no one on – OPS against: .753 (about what Joe Blanton does against everyone)
    With men on – OPS against: .650 (about what Roy Halladay did against everyone in 2009)
    With RISP – OPS Against: .479 (Better than Pedro Martinez’s best year!)

    So what is more likely, that Happ pitches like an untouchable superman with RISP, and Joe Blanton with no one on, or that he was incredibly lucky with men on in general, and with RISP in particular?

    If the answer is that he bears down with RISP, then for pete’s sake why doesn’t he pitch like that all the time!??!!???!??!? He’d be the best pitcher in the history of baseball if he could pull that off!

    See the source material here: www.baseball-reference.com/players/split.cgi?n1=happja01&year=2009&t=p

  6. LarryM

    January 27, 2010 04:16 PM

    Okay, I’ll give it a go .. though not sure I buy this either:

    DIPS Metrics are better predictors of future success than ERA. But like any statistic, they aren’t perfect. Many players will exceed DIPS based predictions. It’s possible that deviations from DIPS predicted performance are merely random luck – but it is also possible that there are factors which are hard to identify statistically which make certain pitchers “better” than their DIPS suggests. For example, some pitchers really do, throughout a career, manage to have below average BABIP numbers.

    Perhaps the people touting Happ are seeing real factors which will make him one of those players.

    Seperate question – I know the studies show that DIPS are better predictors than ERA. But in situations where there is a disparity between DIPS and ERA, has regression analysis demonstrated that the ERA is ENTIRELY irrelevant? In other words, if a pitcher has an ERA of 3.50, and DIPS predicting an ERA of 4.50, we know that 4.50 is more likely that 3.50 next year. But under such circumstances, is 4.50 the “best” predictor, or is (to just throw out a hypothetical number) 4.25 (weighing DIPS 75%) a better predictor?

  7. Bill Baer

    January 27, 2010 08:53 PM

    Larry, sure there are exceptions to every rule. And Happ could be one such exception. However, given the information available at present, we have no reason to conclude that Happ will be such an exception.

    As an example, I went to Baseball Reference and took every pitcher in 2008 who had a BABIP as low as Happ’s (.270) and compared that to their BABIP the following year. The results:

    Armando Galarraga: .237 –> .298
    David Bush: .238 –> .327
    Tim Wakefield: .240 –> .295
    Greg Smith: .258 –> Did not play
    Gavin Floyd: .259 –> .285
    Daisuke Matsuzaka: .260 –> .382
    Jeremy Guthrie: .261 –> .287
    Scott Olsen: .261 –> .348
    Cole Hamels: .262 –> .321
    Todd Wellemeyer: .267 –> .357
    Joe Saunders: .269 –> .289

    The smallest “regressions” came from Gavin Floyd, Jeremy Guthrie, and Joe Saunders. However, the year-after BABIP’s all returned to within one standard deviation (in the .020’s if I recall correctly; too lazy to do the math right now) of the National League average.

    as regression analysis demonstrated that the ERA is ENTIRELY irrelevant?

    I don’t recall reading any studies about that. I’ll look around though.

Next ArticleBDD: A Not So Happ-y 2010