A few thoughts:
- Bill James started the ball of modern statistical analysis rolling with his work in baseball analysis, which led eventually to Moneyball. I guess everybody knows that story by now.
- James' entire project sprung from the insight that baseball is a fundamentally individual sport. By far, the greater part of the competition is determined by the individual performances of pitchers and batters. It should therefore be possible, with enough data and rigorous analysis, to seperate team performance from individual statistics - that is, to determine how many Wins each player is producing. This was the genesis of the Win Shares theory of statistical analysis. The idea was that traditional statistics were worse at measuring player performance than what could be derived.
- Basketball is maybe the single worst team sport for application of this type of analysis because of how much time the starters spend together on the court. Seperating individual contributions with so much overlapping time is a statistical nightmare. The jury is still very much out as to whether the derived statistics are more valuable than the traditional ones.
- Applied to the projection of college performances to the pros, modern statistical analysis is even more questionable. The guys at Wages of Wins have only been doing actual predictions since the 2010 draft, so it is in many ways too early to tell, but we can say a few things about their projections for the 2010 draft. We'll use their preferred statistic, WS/48, to make our comparisons, with a minimum of 1500 minutes played: 2010 Draft Projections:
Actual Draft Results/Production: http://www.basketball-reference.com/draft/NBA_2010.html
A quick rundown of where WoW's projections has done substantially better or worst than actual NBA front offices since the 2010 draft.
- Ed Davis: WoW rank 4th. WS/48 rank 3rd. Drafted 13th. Age at draft: 21.
- Landry Fields: WoW rank 8th. WS/48 rank 6th. Drafted 39th. Age at draft: 21 (three days from 22nd birthday)
- Quincy Pondexter: WoW rank 18th. WS/48 rank 9th. Drafted 26th. Age at draft: 22.
- Wesley Johnson: WoW rank 19th. WS/48 rank 19th. Drafted 4th. Age at draft: 22 (two weeks from 23rd birthday)
- John Wall: WoW rank 32nd. WS/48 rank 16th. Drafted 1st. Age at draft: 19. (Yeah, I know Wall's career WS/48 isn't great yet, but seriously...32nd?!)
- Greg Monroe: WoW rank 20th. WS/48 rank 1st. Drafted 7th. Age at draft: 20.
- Paul George: WoW rank 21st. WS/48 rank 2nd. Drafted 10th. Age at draft: 20.
- Gordon Heyward: WoW rank 16th. WS/48 rank 7th. Drafted 9th. Age at draft: 20.
- Patrick Patterson: WoW rank 25th. WS/48 rank 8th. Drafted 14th. Age at draft: 21.
- Kevin Seraphin: WoW rank 52nd. WS/48 rank 10th. Drafted 17th. Age at draft: 20.
- Avery Bradley: WoW rank 51st. WS/48 rank 12th. Drafted 19th. Age at draft: 19.
- Greivis Vasquez: WoW rank 39th. WS/48 rank 16th. Drafted 28th. Age at draft: 23.
- Eric Bledsoe: WoW rank 50th. WS/48 rank 20th. Drafted 18th. Age at draft: 20.
What can we say from this? Well, first, it is a very limited study. We'd need another ten years to properly assess how well WoW's long-term projections do for these players. That being said, WoW explicitly states that their projections are meant only over the life of the rookie contract - the first four years. So looking at their performance after two seasons is not so out of line. It seems that WoW's model does quite well for upperclassmen. With players 21 and over, WoW scores big wins over NBA front offices four times, and misses twice. With players 20 and younger, WoW does very poorly vs. NBA front-offices, and is especially miserable inside the top-10.
Guess who WoW (and pretty much all of the modern analytical tools) hates this year: Yeah...Harrison Barnes. Food for thought.