Which player would you rather have?

[DSMok1]

Well, they are never perfectly predictive, but they certainly tell you SOMETHING about the true talent level of a player or team, and surely some are more predictive than others. We know, through research, that a baseball team's Pythagorean Record is more predictive of a baseball team's ability than their actual record. I admit that as much as I consider myself a fan of sabermetrics in baseball that I am clueless/skeptical when it comes to hoops, especially college hoops where the variance in abilities is really large. But I am open-minded, and appreciate DSMok1's response.

Here are a few places to start learning about basketball's advanced statistics:

APBRmetrics Forum -- The place to go for advanced stats. People like John Hollinger, Dean Oliver, and others post on this forum. And the predecessor Yahoo Group has good reference material.

Ken Pomeroy's site. The place to go for advanced stats of the college game. Read his blog for tons of good information about advanced basketball stats.

Basketball-Reference.com. All the NBA stats you'll ever need, and the blog has tons of research presented. That's where the latest stuff on Statistical Plus/Minus was done.

Basketball Prospectus -- Advanced basketball statistics.

82Games.com -- NBA statistics and lots of articles about basketball statistical research.

Individual sites:
Journal of Basketball Statistics -- Dean Oliver's early work.
Count the Basket -- Eli Witus' early work. (Both of these guys got hired by NBA teams.)
Basketball-Statistics.com -- Jon Nichols' work; some NCAA stuff recently (including +/- data for college!)
Basketball Geek -- Ryan Parker's research; some new good stuff on college basketball.
Hoopdata -- Advanced box scores and a blog, for the NBA.

 

[DSMok1]

Well, they are never perfectly predictive, but they certainly tell you SOMETHING about the true talent level of a player or team, and surely some are more predictive than others. We know, through research, that a baseball team's Pythagorean Record is more predictive of a baseball team's ability than their actual record. I admit that as much as I consider myself a fan of sabermetrics in baseball that I am clueless/skeptical when it comes to hoops, especially college hoops where the variance in abilities is really large. But I am open-minded, and appreciate DSMok1's response.

Some statistics are predictive, yes. Baseball lends itself to these things (for instance FIP predicting future ERA far better than current ERA does). However, in baseball it is possible to measure each player's impact separately, and so to get a very precise estimate of each player's ability and value. Basketball has 5 players on each team at a time; it is very difficult to pin down each player's value. Nay, I would say impossible. All we can do is approximate. (Of course, it is even harder in football with 11 x 11.) Since we barely even know what players are actually worth, we are not too great with predictive metrics yet. Basketball requires a blend of scouting and statistics; baseball can be analyzed almost purely on statistics, and football requires mostly scouting (though we can learn about skill-position offensive players at least somewhat via stats.)