Where The Value Lies

,
I'm going to start running these posts on Monday mornings, in lieu of the normal leans posts. At first, they are going to be rather elementary. I really don't know a ton about baseball (which should be obvious by now), but it seems to me, the most basic way to determine whether a team is overrated or underrated is the look at the difference between teams' actual records and their third-order pythagorean records from BP. It corrects for strength-of-schedule and predicts the number of runs that should be scored from teams' peripheral statistics. These are as of games from May 9.

Overrated
Red Sox +2.9
Giants +2.7
Marlins +2.2
Reds +2.2
Brewers +1.9
Dodgers +1.9

Underrated
Indians -4.7
Nationals -3.8
Rockies -3.2
Rays -2.8
Diamondbacks -2.4

The underrated teams really don't surprise me. Washington, Cleveland, Colorado and Arizona have all been blog favorites. I'm somewhat surprised to see the Reds and Brewers on the overrated list, though.

As I said, I don't really have a wealth of baseball knowledge, so ideas on ways of improving these posts are welcome.

Streak for the Cash
10a Deccan @ Rajasthan (Indian Premier League - Cricket)
Current Streak: 2

With today's options being an either/or, Deccan is the best choice. Hopefully, it is done by 7p.

11 comments:

Jim D. said...

Don't some cricket matches take days?

LMQ said...

I'd be interested to whether and by how much the Indians would be underrated if you removed the 22-4 Yankees game.

am19psu said...

I'd be interested to whether and by how much the Indians would be underrated if you removed the 22-4 Yankees game.

Looking at first order records, the Indians projected winning percentage falls from .423 to .364. Their first order difference falls from -2.6 to -1.3, so they are still underrated, but they would be the 6th most underrated team.

Frankly, I don't have the time to figure out the adjusted equivalent runs to get the third order difference.

Of course, this whole exercise is pointless anyway, since you just can't take points out of a sample without reason. Yes, Wang was hurt, but that 22-4 beatdown is part of the Indians' resume. It's not like we (or computer rankings, to be more analogous) discount Texas Tech's drubbing by Oklahoma because the rest of their resume looked so good.

rexfordbuzzsaw said...

As far as second order runs go, they would still be about the same. Yes, 20 runs is not that common, but the Indians had a ton of baserunners/total bases that day, so its not like they didn't deserve to score a ton of runs. Without going through my record, I'm pretty sure the Indians had over 20 using the baseruns formula, which means that game is pretty irrelevant to their over/underratedness.

am19psu said...

Yes, 20 runs is not that common, but the Indians had a ton of baserunners/total bases that day, so its not like they didn't deserve to score a ton of runs.

I agree, but I think LMQ wants that game removed the record entirely, so they wouldn't get credit for any of the baserunners/EqR/baseruns/etc.

Vegas Watch said...

"I think LMQ wants that game removed the record entirely, so they wouldn't get credit for any of the baserunners/EqR/baseruns/etc."

Which is, of course, going to give you a much less realistic depiction of how they've played than if you included that game.

LMQ said...

I didn't say I wanted it removed from the record (or really anything close to that). I was really just curious how much that one game affected their "underratedness."

Hypothetically, say that removing that one game moved Cleveland from being underrated to being overrated. Don't you think that piece of information would be worth knowing?

am19psu said...

Hypothetically, say that removing that one game moved Cleveland from being underrated to being overrated. Don't you think that piece of information would be worth knowing?

Only if removing the game gets the calculation closer to the true expected value, which is impossible to know without a population or having a legitimate reason for removing the point.

In any case, I have to imagine, given the calculation for the first order record, that the point is moot.

am19psu said...

Look at it another way: you can roll '6' a bunch of times in a row on a fair die. Also, there is a probability that the Indians could have strung together that many hits/walks in a row to score 22 in a game. You wouldn't remove the sample of rolling fifteen '6's in a row just because it was improbable when calculating the EV of the die.

Does that make more sense?

LMQ said...

What you're saying makes perfect sense and I completely agree that there is no reason to exclude that game.

I probably wasn't as direct as I should have been earlier, but my real point is that the smaller the sample of games, the less meaningful the results. I guess my earlier question should have been more along the lines of "how much does removing ANY one game affect a team's under/overratedness?"

To put it another way, I'm saying that the information presented in this post is more valuable now, after a month plus of games than if the same study had been done a week into the season.

LMQ said...

Oh, and I should also say to keep up the good work. I really enjoy reading your stuff here.