Yeah, 100 is better than 99. Real tough stuff. But do you really think you "perfectly" understand ratings and how they work in conjunction with all the other moving parts of HBD?

Because I don't think you do. I think I'm in the best worlds in HBD. I'd compare them, favorably, to Gleeman worlds. I'd even imagine I understand the ratings better than you. And I'd say I'm very far from having a "perfect" understanding of them.

Fine perfect was a bad word choice.

If you have 90% knowledge from ratings the statistical difference required to say these guys fall in that 10% is impossible to occur without seeing it in the ratings.

So you don't have 'perfect' knowledge. But you have so much (again if you've done the math) that stats shouldn't play any part in an evaluation.

Stats win the games. If you accept that you don't have a perfect understanding of the ratings, factoring in the stats that stare you in the face might be a good idea.

Which guy do you take?
66/74/57/53/48 or 72/65/53/54/51

I honestly can't say. Right now, I give one a try and, if his STATS say "Not really cutting it", I give the other guy a shot. Spring training, as I suggested, would help me make the decision before the games count. Thus, spring training would have some meaning. It doesn't have any meaning now. So why the disagreement?

If you think you can get any read on which of those two guys is better by even a full season of stats, you're crazy.

They are so close you could flip a coin. Stats aren't going to tell you anything the ratings don't.

You'd be better off flipping a coin that waiting for even a full season of stats to give you any actionable knowledge.

And if you give the first guy a shot and he's good, how do you know the second guy wouldn't have been better. Just play the better fielder.

Just pick one and trade the other if you don't need them both. At least then you don't tie up a roster spot for a year.

"Stats win the games."

Uh no they don't. Stats tell you who won the game. The engine is driven by ratings not stats. This isn't rocket science.

Quote: Originally posted by MikeT23 on 5/20/2010Which guy do you take?
66/74/57/53/48 or 72/65/53/54/51

If the two players are close enough that you can't immediately tell which is better, then they'll perform within the margin of error of each other. It's possible that your or my knowledge (especially mine, I'm ok at statistics but still new at HBD) of how ratings get turned into statistics is limited enough that there actually is a significant difference between these two players. Even if there is a difference, though, you won't be able to determine it statistically in 18 games.

So the answer to your question is: pick one. Random selection in this case will be as effective as any decision based on 18 games worth of data.

Quote: Originally posted by saintonan on 5/20/2010

Quote: Originally posted by MikeT23 on 5/20/2010Which guy do you take?66/74/57/53/48 or 72/65/53/54/51

If the two players are close enough that you can't immediately tell which is better, then they'll perform within the margin of error of each other. It's possible that your or my knowledge (especially mine, I'm ok at statistics but still new at HBD) of how ratings get turned into statistics is limited enough that there actually is a significant difference between these two players. Even if there is a difference, though, you won't be able to determine it statistically in 18 games.

So the answer to your question is: pick one. Random selection in this case will be as effective as any decision based on 18 games worth of data.

He said what I was trying to say better. Thanks.

Quote: Originally posted by goosegoslin on 5/20/2010"Stats win the games."

Uh no they don't. Stats tell you who won the game. The engine is driven by ratings not stats. This isn't rocket science.

Ah, ya - they do.

If one guy is 75 power, and the other is 70 power, you know unequivocally that the 75 guy has more power.

If however we were on a 2-8 scale, with +/- 1 point either way, the 75 and 70 would both show as "6" power with perfect scouting (despite the underlying 70 and 75). But either could be shown to a particular user as between 5 and 7, depending on your scouting budget, etc..

Then you could have a case where a 5 truly had better power than a guy your scouts say is a 7.

If that were the case, than yes, looking at stats could have a big impact in helping you find cases where the scouts are wrong.

But with the current HBD setup, there's no need to for that, in fact it's counter productive. You know the guy with 75 is better, no matter what their stats are. Any statistical difference is luck or environment (which pitchers a guy faced, which park he played in, random chance, etc.).

And if you've crunched the numbers, you can get a pretty good formula to combine the ratings on offense, until they go and de-steroid the engine or something. Then you wait for enough new data and tweak your formula.

Someone posted just the other day that they have an offensive rating system that pretty much ranks player dead on (when compared to OPS) given a big enough sample size. That would be MUCH harder to do, with a lot more grey area if we didn't have perfect knowledge of current ratings.

And by perfect knowledge, I don't mean that we're able to combine the ratings perfectly. I mean that we know, 100% for sure that a 78 is better than a 77. There is no 'fuzziness', like there is with future ratings.

Quote: Originally posted by mitchrapp on 5/20/2010

Quote: Originally posted by goosegoslin on 5/20/2010"Stats win the games."
Uh no they don't. Stats tell you who won the game. The engine is driven by ratings not stats. This isn't rocket science.

Ah, ya - they do.

What are you talking about? Stats haven't ever won a game. Everything players do is based on their ratings. This isn't a tough concept.

Apparently it is if you believe the guy who bats .400 doesn't help you win more games than the guy who bats .250 but has much better ratings.

This really isn't a tough concept.

I must admit I didn't read all that nonsense but "flip a coin" is a fairly stupid way to decide who the better player is if the opportunity to look at some meaningful stats present themselves.

This, too, really isn't a tough concept.

And, for the record, the score tells you who won the game.

The guy hit .400 because of his ratings and luck (or unluck, if his true ability is over .400 somehow).

There are not stats that caused him to hit .400. It his ratings, his opponents, the park and a random number generator. That's it. No stats. Stats tell you what happened. They don't decide what happens.

What I see is a couple of people "OH NO!!! I DON'T WANT ANY ADDITIONAL INFO WHEN I MAKE ROSTER DECISIONS!!!! I'LL JUST LOOK AT THE RATINGS AND FLIP A COIN IF I CAN'T DECIDE!!!"

Dumb.

Quote: Originally posted by MikeT23 on 5/20/2010And, for the record, the score tells you who won the game.  

The score is a stat. You are firing on all cylinders today.