Oakland A's Performance Shows That 'Moneyball' Doesn't Always Pay Off

[nny]

uhh now you're really confusing me. It does use game data. I don't know how I implied it didn't.

I think we're talking about two different things lol

I think the confusion is over the "hitters aren't responsible for the situation" thing I brought up.

What you were talking about, or at least what I got from what you were talking about, was looking at each individual event and saying how much is a run worth there. For instance, a single with a runner on second is worth more than a single with nobody on. I believe that is what WPA does. But if Player A and Player B put up the same exact 1b/2b/3b/so forth, and Player A is on the NYY and Player B is on the Padres, that makes Player A look more valuable than Player B. Yes, Player A was responsible for more runs, but if both are free agents that's no indication of future success on his new team.

The stat I'm using look at the worth of all singles and gives what the average single will be worth based off regression.

And any time you talk about value in baseball it's theoretical really, but now I assume we were talking about different parts being theoretical since you thought the stat didn't use game data lol.

[CrimsonCane]

I think we're getting confused over your use of the word "theoretical." I'm using "theoretical" in contrast to "applied." Baseball statistics are not theoretical, they are applied. Theoretical statistics rely on values determined without the need to aggregate real world data. You don't need to roll a dice 1,000,000 times to determine that you have a 1/6 chance to roll any particular number. You can use simple math for that. That's relying on probability "theory". Whenever you take an average, you are doing "applied" math. No theory at work. Theoretical does not simply mean that something has been assigned an average value instead of a more context-specific one.

And, as for relying upon more situationally specific values, I was a bit confusing at first. My main point, which I buried, was that steals should be weighed based upon the in-game situation in which they occurred. I'd agree that by and large it's unfair to take hitting stats (1B, 2B, 3B, HR) and assign value because the single happened in one situation for Player X and another situation for Player Y. However, I don't think it's unfair to do that for stolen bases. Base running relies significantly more upon an player's ability to read in-game situations and make decisions based upon those observations.

[nny]

And, as for relying upon more situationally specific values, I was a bit confusing at first. My main point, which I buried, was that steals should be weighed based upon the in-game situation in which they occurred. I'd agree that by and large it's unfair to take hitting stats (1B, 2B, 3B, HR) and assign value because the single happened in one situation for Player X and another situation for Player Y. However, I don't think it's unfair to do that for stolen bases. Base running relies significantly more upon an player's ability to read in-game situations and make decisions based upon those observations.

let's look at these two situations.

Bottom of the 9th, runner on first, no outs, 9-1 game.
Bottom of the 9th, runner on first, no outs, 2-1 game.
Both situations, the runner than steal's first.

Regardless of the score, both situations adds a flat rate to the chances the home team now scores a run. That we both agree with yes?

Now you're argument is that the second scenario, the SB means a hell of a lot more to the team's chance to win. Yes?

For that, you could add an leverage index to it. That's what FanGraphs does to relievers with their WAR stat (after all, 1 run given up in a one run game is big, a run given up in a blow out is not, so it credits the guys used it big situations and discredits those used in blow outs).

Originally I was going to say I disagree, but honestly I do agree. I think the LI for relievers is a very smart thing. But ontop of that I have really been wanting some kind of stat to do that with hitters. Not to base them off who's on base and who's not, but more so who's the pitcher on the mound. When a game is 9-1 and you hit some jack against some positional player pitching for the first time since high school, that's far from hitting a jack in a 2-1 game against Mo.

What I'm more interested in is who are the offensive stat padders. It probably doesn't mean that much of a difference in the regular season, but I imagine if you "build for the playoffs" it really does matter. I actually imagine that would be the "next step" as Swift put it. I mean there are several hitters that get the "bad ball hitter" labels, I want to know who those hitters are and objectively say how much that impacts their overall ability.

[lou]

Oakland has a + 16 run differential with no regular over a .790 OPS (Holliday is over it for his 350 or so PA) and currently 111 starts to rookie pitchers. In fact, I'd say this year is a shining example of how much it does work with the team they are fielding.

Fully expect the natural swing of "Beane is genius" articles when they move to a + 75 next year when the offense and pitching undoubtedly improve and the W/L is better reflected by their actual scoring/giving up runs performance.