Advanced Statistics Sites

[mogwai316]

http://www.popcornmachine.net/

[Pudge714]

http://hockeystats.no-ip.org:81/nhl.php?id=0

[Jack of Arcades]

http://dberri.wordpress.com

[Pudge714]

JoA,
I have heard lots of bad things about Wages of Wins, someone posted a really good critique of it at 82games, I assume you think it has gotten, better. Isn't his method pretty flawed since it looks at VOAP instead of VORP?

[MikeyPatriot]

www.fightmetric.com (not a Basroll)

It's a new analysis site about MMA.

[Jack of Arcades]

[ QUOTE ]
JoA,
I have heard lots of bad things about Wages of Wins, someone posted a really good critique of it at 82games, I assume you think it has gotten, better. Isn't his method pretty flawed since it looks at VOAP instead of VORP?

[/ QUOTE ]

I'm just putting it out there. I don't think looking at average instead of replacement player is too bad, I mean, tons of stats do that. I'd like to see the critique though.

[J.R.]

The best resource for baseball analysis is The Book Blog.

[kidcolin]

JoA, Pudge,

I too have read some critiques of Wages of Wins. In fact, a couple months back one of the big stat dudes from APBR metrics (or associated with it or whatever) was presenting a paper somewhere in Boston slamming it. Basically saying it isn't sound science.

I know one of the critiques is there are no position adjustments. I've also read about a "fudge" factor, but I forget what it is exactly.

All that being said, I still read his stuff and a lot of it is pretty interesting. It's also interesting that he's been pretty accurate with his win-loss predictions in the NBA. Plus he loves Pierce so he's obviously not all bad.

[sethypooh21]

[ QUOTE ]
http://dberri.wordpress.com

[/ QUOTE ]

LOL.

[sethypooh21]

Several major problems with WoW (my critiques, though APBR probably has more. There was a huge knockdown dragout between Berri, Hollinger and the dude who works for the Sonics about 18 months ago):

1) The "fudge" factor by which stuff is scaled towards a teams actual results. Inherent is the assumption that everything that goes into winning is captured by the numbers input into the formula. Though this assumption makes some sense in Baseball were ABs can fairly be described as independent random events, there is too much context in a basketball game for anyone who knows anything about basketball as played rather than on paper to feel comfortable making that assumption about anything but FTs (and even then, ignoring fatigue and game situation factors is a leap which to this point is unsupported)

2) MASSIVE overvaluation of rebounding from an individual standpoint. Berri notes that rebounding is important to team success. The problem comes from reducing the team metrics to individual player values - the player who actually gets the defensive rebound gets sole credit for gaining possession when presumably the man defending the shooter and the other players boxing out probably had something to do with it. This leads to absurd results such as bangers who never shoot anything but dunks being overvalued pretty significantly.

3) Since the individual stats are tied to team performance (and team performance is subject to many factors not especially well measured by convention stats) WoW has virtually no predictive power (a point which Berri at times agrees, yet at other times chooses to conveniently ignore.)

And the overriding criticism is that Berri demonstrates no understanding of how a basketball game occurs in real life rather than on paper. Again, the difference between baseball (which translate exceptionally well to statistics, with the possible exception of defense, and those metrics are getting better all the time) and basketball is fairly striking. But DB has no cognizance of this, or at least has not demonstrated this.