After Sabre-School Special (June 19, 2003)
A followup
--posted by TangoTiger at 03:42 PM EDT
Posted 3:43 p.m.,
June 19, 2003
(#1) -
David Smyth
These comments are for the After Sabre School thread, but there is nowhere to post a reply there.
Bill James does the following in settling on a PF for, say, 1998:
1996, 1/8
1997, 1/8
1998, 1/2
1999, 1/8
2000, 1/8
This a bit different than just using the 5 prior seasons. He is weighting the indicated season higher than the others to account for weather differences and other real year-to-year variations. This is reasonable since he is not using actual weather info. Maybe the weight should be less than 1/2, but aside from that, I think his structure is fine.
And as far as the Pythagopat, I thought it was settled, from your little study, to use .28 (not .29)?
Posted 3:46 p.m.,
June 19, 2003
(#2) -
tangotiger
Park: well, that makes alot more sense. I should have verified what he did. That looks just about what you and I discussed a year or two ago, about putting the PF at the center of the time period.
28/29: well, .28 is my preference, because I look at extreme teams. .29 is the best-fit for actual all teams.
Posted 5:42 p.m.,
June 19, 2003
(#3) -
David Smyth
As far as using the B James Theoretical Team concept with BaseRuns:
The usual procedure is to substitute the player into an avg team. But the more correct concept (I believe) is to substitute the player into a "random" team. It may seem, at first blush, that this amounts to the same thing. But maybe not. If you put 2 players on an avg team, who both have the same runs added, but 1 has a hi SLG profile, and the other a hi OBA profile, they might NOT have the same impact added to a random team. And that is because SLG (advancement) has pretty much the same value in every realistic context, but the value of baserunners varies more. Therefore, I propose that a hi OBA player is actually more desireable than a hi SLG player, even if they seem to have the same value added to an avg team. I am going to compute some examples of this, but I will be surprised if it doesn't hold up. And if it does, that means that the proper measure of a hitter's ability lies somewhere between Theo. Team BsR and pure Bsr>
Posted 7:37 p.m.,
June 19, 2003
(#4) -
Patriot
I'm not sure I quite understand the idea of a random team. I know that Wolverton does this in his Pennants method as well.
If you put the player on every possible random team, and then weighted the player's performance by the percentage of actual teams that perform at that level, would not the result be almost the exact same thing as just sticking him on the average team? Maybe I am missing the boat here, I'm not debating, I'm asking.
OTOH, I do see what you're saying about OBA v SLG. A higher OBA is pretty much always a good tradeoff.
Posted 7:47 p.m.,
June 19, 2003
(#5) -
tangotiger
I agree with Patriot that I don't think you'll see any difference (max 1 run / 600PA is my guess), especially since a random team really isn't that different from the average. A team .300 to .360 OBA is no big diff really.
Posted 7:48 p.m.,
June 19, 2003
(#6) -
tangotiger
I mean no big diff for what's being proposed here.
Posted 8:18 p.m.,
June 19, 2003
(#7) -
David Smyth
Well, I didn't say that the difference will be "big". I just said that it should be "systematically real", and as big as a lot of the other minor differences that people like Tango, MGL, Patriot, try to elucidate and quantify. Why is this concept apparently being held to a higher standard of significance?
Posted 8:45 p.m.,
June 19, 2003
(#8) -
Patriot
Don't get me wrong, I say go for it and see what you find. See if the added complexity is worth the extra work, and if it is, then great.
Posted 8:56 p.m.,
June 19, 2003
(#9) -
Vinay Kumar
I agree with David that it makes sense to look at the random team, while also agreeing with Tango and Patriot that it probably doesn't make a noticeable difference.
However, Wolverton's Pennants Added is very different. When it comes to winning a pennant, the random team is very different from the average team, because a pennant is an extreme performance, so the distribution matters more. I'm pretty sure you guys understand that, but I just want to clarify that that's an important distinction.
Posted 9:15 a.m.,
June 20, 2003
(#10) -
Warren
I just wanted to follow up on what Vinay said - he's exactly right that when you're looking at pennants added, you get a different picture. One thing I thought about when playing around with this pennants added stuff is that we shouldn't necessarily come up with a single baseline (average, replacement, random), but rather look at *every* baseline. A player with a very high peak is more valuable to a poor or average team, but a consistent player is more valuable to a great team. There's no right or wrong answer as to who the better player is - like many things, it depends on the environment.
One other baseline, and one more important from a value perspective, is his actual team's performance (subtracting out himself, I suppose). It's certainly possible to have two players that would theoretically have equal value on a average or random team, but that have additional value because of the types of teams they play for (a high peak player on a bad team, or a consistent player on a great team).