This is a thread on which we discuss baseball without being argumentative douchebags. Will it attract any posters?

Just everyone stop trying to be the biggest asshole on the board, okay, it's not junior high school.

Is there an inherent limitation in statistics-based baseball analysis? This can also be framed the other way: Why, exactly, should there be something truly ineffable in baseball or sports that defies numerical and historical trend analysis?

Well, sure, defense! But they're working on that. Seriously though, things such as chemistry and makeup of course cannot be quantified -- nor can the HUMAN SPIRIT -- but those are impossible to gauge in any sort of objective manner anyway, so that is hardly SABR folks' fault.

JESUS BILLY, THEYVE SOLVED BASEBALL...

On the other hand, while Joe Morgan was a wonderful and saberfriendly baseball player, his absolute refusal to admit any stat-credibility whatsoever smacks of a little kid putting his hands over his ears when he's told that Santa Claus isn't real.

So I'm not on anyone's "side," officially, at least here. I played baseball for a few years, but the only things I was good at were drawing walks (SABR), base-stealing (OLD SCHOOL), and infield chatter, except really outfield chatter because that's where I was banished. I can very easily see how insular baseball coach-player culture wouldn't think they needed any outsiders telling them what to do, kind of like hillbillies with moonshine stills wary of the revenuers. But why CAN'T we give a team IN A PARTICULAR YEAR credit for a huge edge in one-run wins? Isn't it just that there's no algorithm for it yet? And, if that's the case, why is it that statheads are all like "nope it's all luck it's all luck no credit"?

i wish you didn't hate me john, i've stuck up for you before, but whatever

Regarding players and doling out MVP awards, we had this discussion a few weeks ago. Start with Milo's comment here:

2005 Awards (MVP/Cy Young/ROY/etc) Speculation Thread

A few posters argue for exactly what you're suggesting -- we can give a guy credit for being TEH MAN on a first-place club and driving in big runs in important games, even though luck is involved with placing TEH MAN in so many of those key situations.

and to g!: on that other thread, to cite one example, i think hstencil is (unfairly) characterizing barry in that way.

Yeah, it proves that Pujols is 100% CLUTCH.

i think hstencil is (unfairly) characterizing barry in that way.

yes, but Barry handled himself in a very impressive manner (which adds to his charm and character, esp. as a mod). this is an internet message board and undoubtedly someone is going to overstep the bounds of common courtesy and tact at times, but hopefully the good will overshadow the bad in the end.

(G!: I didn't realize that this was a referendum on Barry's demeanor, but of course I agree with you.)

.125/.200/.325

Pujols' line in "close and late" situations:

.250/.365/.853

Batters facing Lidge in situations with men on second and third:

.048/.167/.214

Lidge after throwing more than 16 pitches in an inning:

.160/.232/.512

Lidge facing batters with two out:

.105/.209/.420

Which goes to show not Pujols' CLUTCH factor, but that that was an extremely fortunate shot. The odds were very much against him.

Yeah, isn't this similar to looking for definitive proof that a particular player in a particular year (say, David Ortiz) was only "lucky" to have so many game-tying or game-winning hits?

I'm not sure what you're getting at with "it's only probability"? The probability that Pujols homered off Lidge last night was damn small. The probablility that any other hitter did the same thing in the same situation would have been hell of a lot smaller. No matter who was hitting, the Cards were lucky last night, because the chances of any team winning a game when down by two with two outs is incredibly small.

I don't know that it's statheads saying IT'S ALL LUCK YOU SUCK as much as it's statty folk (over)compensating for the "they know how to win" malarkey by emphasizing the statistical anomoly of performances such as:

CHI 35-19 (best in the majors - +16) (+37 overall)
TOR 16-31 (worst in the majors - -15) (-2 overall)

Credit's due to the White Sox for being a very good team (with ridiculously great pitching and oddly opportunistic hitting), but being a very good team != being good in one-run games. Tampa Bay is +4 in ORGs; St. Louis is -4. COLORADO was 25-24 in ORGs. Talent helps (duh), but the circumstances that need to come about for a team to win / lose a one-run game are often just as random & unpredictable as the circumstances that cause the one-run game to happen in the first place.

what it means is that Dustin Hermanson is a CLUTCH pitcher.

i believe in the mathematical AND the ineffable, i believe there is much at work that we don't know and can NEVER predict fully through either science or faith, i believe that the novels of susan sontag ARE actually pretty good but that long slow deep wet kisses that last for three days are kind of disgusting, etc.

No, it's random because there's no correlation between winning and winning ORG's. Randomness can be rigourously defined (more or less).

OTOH, you can argue that we shouldn't evaluate how lucky a team was by looking at their records in one-run games -- maybe there is *another* mathematical formula that is much better suited to that task!

and how fucked-up am I for putting "good" in quotation marks up there?

This was more of a culturally-developed habit, rather than a serious attempt to understand what makes a great hitter a great hitter. But point taken.

what if someone figured out the fractal order in that apparent randomness?

Yes, exactly -- we could argue that *all* ORG's shouldn't be considered in the argument. Instead, we could only consider the ORG's that were won in the 7th inning or later. Or the ORG's where starters left with a lead and the bullpen protected it. Or any number of other possibilities.

But the more you restrict the data in order to look for meaningful correlations, the smaller the sample sizes become and the more unsure we are that our conclusions are statistically significant. There is no way around this. However, it doesn't mean that we can't look at small samples, but we need to be aware of the larger uncertainty in the conclusions we reach.

"The Cardinals made a couple of terrific defensive plays in this game that kept them alive. Yadier Molina's second-inning catch-and-tag that retired Morgan Ensberg was as good a play as a catcher can make. It wasn't just the physical act, but the split-second decision by Molina to turn to his right, rather than his left, converting the bad throw by Albert Pujols into an out. Just a great play, another one of those elements of defense that we have no real way of measuring."

I don't remember this in Moneyball (tho it could well be in there) but Alan Schwarz covers it (and Alexander Cartwright in general) in that "Numbers" book from earlier this year.

"After making quick work of John Rodriguez and John Mabry, Lidge had driven the Cardinals' chances to just 1.2%. Eckstein's two-strike single (2.4%) and Edmonds five-pitch walk (3.6%) crept them back to 7.3%. Normally, I would think that 7.3% in that situation is generous, especially with a pitcher the caliber of Lidge on the mound. However, with Pujols at bat that probably was about right.

The 0-1 hanging slider, however, that would have been a homerun in any park registered on the win probability chart like a seismic event, raising the Cardinals win probability an amazing 73% to 80.3%. Jason Isringhausen then closed the door in the 9th and put the Astros in a "state of Missouri" (ok, not original but still funny right?)."

For instance, according to my knowledge of statistical theory, in the moment where t (time) approaches zero, the application of statistical probabilities becomes negligible -- how does the 30% chance of a hitter making a hit manifest itself into a single at bat (ignoring for the moment the luck-driven nature of the "hit")? Of course, from a more epistemologically divorced standpoint, i.e. from an angel's eye view, these probabilities take on more material shape.

But what good does an angel's perspective have on us mere mortals? We live in the day-to-day, in the moment where time becomes statistically negligible. What are the odds that something happens in a given moment? They either do, or they don't -- 50-50. But withdrawing towards epistemological heaven, where the sample size is more useful, you have greater subtlety and variability.

i don't follow this at all!

Phenomena are only explicable retrospectively, but there is a certain magic in the individual, discrete points from which statistics are gathered. I assume the two camps in baseball criticism and theory occupy, respectively, the metaphysical and statistical positions, a debate that has raged ever since the invention/discovery of Science and its kin, Rationalism.

Datta. Dayadhvam. Damyata.
Shantih shantih shantih