The Problem with Sabermetrics

I came across an interesting read about sabermetrics. This article does not necessarily take a stand against them. But I think does a good job of addressing some of the reasons for controversy.

Pro or con I think it is a good read.

http://www.lookoutlanding.com/2011/7/20/2285655/the-problem-with-sabermetrics

Very good read. Recommended! Thanks for sharing it.

This is completely untrue, not to mention lazily ignorant:

"Proper science requires, at the very least, controlled experimentation, which is something impossible to manage in baseball analysis."

In fact, sabr provides many testable hypotheses. Just for one example, it attempts to project future performance. Such attempts can be tested for predictive significance, and therefore they ARE falsifiable, i.e., state a legitimate scientific hypothesis.

xxx

and this is cornball junk that says nothing about the subject:

"Sabermetrics shouldn't be so incomprehensible so as not to call up the smell of fresh mown grass in midsummer, or the crack of the ball off the bat, the blur of seams as an outfielder whips a throw in towards his cutoff man."

Why doesn't he just randomly copy and paste the worst pseudo-lyrical improvisations of Ray Bradbury on an especially hacky day grinding out words that 13-year-olds find poetic?

And that straw man certainly is big.

I think he points out a major flaw in many poorly analyzed tests/experiments/theories...the leaping from association to causation...and without an appropriate grounding in the game, this can lead to flawed conclusions. The example he uses about triples is a perfect example of that.

I wouldn't say that it's a problem inherent with Sabermetrics though, but one that is inherent in flawed testing, and is even more dangerous if the conclusions of those flawed tests are accepted at face value just because they are published as a legitimate test without an appropriate level of critical analysis.

I could go on and on, but that's my 2 cents (for now)...

and that's all I got to say about that.
It is really easy just play against Frank, Visick, Keyzick , Kev and Jeep and it all works in your favor.
If you got Frank and Kev in your division then 2nd is the worst you can do. No math or syberfrank here just Kev it to the bonecracker and you got MC= Hammertime baby!

ESPN magazine had a great article about baseball data -- and WAR, in particular -- earlier this year:
http://espn.go.com/mlb/story/_/id/89595 ... n-magazine

Interesting piece on Sabermetrics but especially so since I analagized it with Economic thoughts and priciples.

As a trained Economist and philosopher I have been fascinated at how poorly my models function as predicters and even worse how often they fail tests as contemporary explanations. Yet, I know my Philosophy to be right!! Am I delusional?

The problem with Sabermetrics(and economics) is that we are dealing with unmeasurable calculaus. The predictions on how Mike Trout will do statistically as opposed to Clint Barmes are predicated on constantly changing parameters-Trout might break a toe and lose an entire season, Barmes might somehow learn to hit. This muddies the science. An Economic example-used sneakers might become worth stealing

The solution is time. These Stratomatic leagues are more valuable statistical tools than real baseball because we play so many more games.(abet they lose a LOT of value due to small amount of teams in the draft pools).Bottom line-100+ years of evolving Baseball is not near enough data to give us total predictive value.

More interesting to me is using Baseball as a tool for Economic system analysis. I can aver defiinitivly that any nationstate be it Cuba or Sweden that adopts a Socialist Economic Base will eventually falter. Sometimes its obvious ala the Pittsburgh Pirates of Economics-Cuba. But Sweden gets a break because their league(the European Common market) is also playing Batting Average over On base so a poorly managed Swedish team might do okay in their league. Of course, evedntually they may lose out to better leagues(the Asian league?).

Also a team might be loaded with good players(say the German team) and can win dispite poor management. The comparisons go on and on and back and forth . It boils down to accurate statistical analysis( or economic truisms) is true but Trout might still break his toe!!

Sabermetrics make for the best debates about baseball because few people use and regard them in a moderated way. Many people either blindly follow the WAR, or believe that old-school analysis of using two eyes is the only way to go. As a whole, I find Sabermetrics incredibly useful. WAR seems to be very very accurate, and advanced fielding metrics such as UZR are waaaay better than errors and other useless basic fielding stats. I think the issue of it being not a controlled experiment is relatively unimportant. All of these statistics should be looked at with a degree of skepticism because the theory behind them is not always rock solid. For example, the commonly used DIPS theory, using research of correlations, asserts that pitchers don't control how hard the ball is hit off them. The first time I saw this I thought it was crazy, and I still think it's wrong. It's widely accepted now, but not unanimously. I came across a great article recently that uses pitch data to assert the opposite.

http://www.baseballprospectus.com/article.php?articleid=15532

I was really glad to see this assertion, because for someone like me who loves the numbers but also uses my eyes, it didn't make any sense for pitchers to have negligible control over batted ball speed.

The whole topic of projections, I think, shouldn't be looked at from the perspective of being a faulty experiment, but rather that the complex formulas, with many notable exceptions, tend to work better than off the cuff projections. The computerized algorithms are more consistent, and although they leave much to be desired in the form of data that cannot be put into statistical form, they are the best there is to date. Anyway, I just felt like sharing my take on sabermetrics and the lack of a balanced approach in evaluating their use.