As noted yesterday, Erich Doerr – via a nifty Monte Carlo simulation — called the NCAA championship game before the tournament started. Although we can’t expect such precision every year (so don’t go betting your life-savings next year on Erich’s picks), it is still neat to see statistical analysis produce such results.
In addition to looking at the NCAA tournament for the WoW Journal, Erich has also written posts examining the NBA draft (here is one from last June). And last year his analysis (and mine as well) indicated that the most productive player taken in the draft out of college last season was Nick Fazekas.
Tables One and Two report the analysis of Fazekas last summer:
Table One: PAWS40 and the Top Prospects
Table Two: Analyzing the 2007 NBA Draft
And here is what I observed last July (I will get to Erich’s insights in a moment):
Topping the list in 2007 is Nevada’s Nick Fazekas. His PAWSmin of 0.269 surpassed Paul Millsap’s mark of 0.244. Fazekas, though, is going to the Dallas Mavericks, a team that already has many of the pieces in place. It’s hard to believe that Fazekas will get much playing time in Dallas next year, so it might be awhile to see if the numbers he posted at Nevada will translate into the NBA. Obviously many NBA scouts have their doubts. Again, college numbers are not a perfect indicator of pro success, so it’s more than possible that the scouts could be right. Then again, they could be wrong (have I covered all the possibilities here?).
As noted, Fazekas was taken by the Dallas Mavericks in the second round. But after playing only nine minutes this season, he was cut from the team. In the last few weeks Fazekas has re-surfaced with the LA Clippers. And although he has only played about 200 minutes, the returns suggest that Erich is once again on to something.
In his first 207 minutes with the Clippers (not counting Tuesday night’s game against Denver where he was once again amazing), Fazekas has posted a 0.316 WP48 [Wins Produced per 48 minutes]. Yes, the sample is quite small. But on a per-minute basis, it might be the best mark of the 2008 rookie class (I haven’t looked at all the rookies yet, but I would be surprised if another rookie surpassed the 0.300 mark).
When we look at the individual statistics we can see where Fazekas has excelled.
Table Three: The Fabulous Fazekas
Table Three reveals that Fazekas is above average with respect to shooting efficiency, rebounds, steals, and blocked shots. He has also avoided turnovers and personal fouls. And when we look at Win Score, not surprisingly we see a mark that’s well above average. In sum, Fazekas has been fabulous.
Of course before we get too excited, let me just note the following.
- Fazekas has only played200 minutes. So our sample is really small.
- Fazekas has not shown that he can perform as part of the regular rotation. That being said, in the seven games he has played at least 15 minutes he has been above average each time.
- And it’s possible that his success has only occurred against the NBA’s lesser talents.
Such a possibility was noted by Erich last summer:
Since Win Score was derived from the evaluation of the NBA, there wasn’t a strong need to adjust for level of competition. The NBA teams sport a much narrower talent gap than the NCAA conferences and International leagues. In looking at college players, though, we have to note that the wider NCAA talent distribution allows for players to pick on the less skilled teams. The following tables offer an assessment of how well various players played against NCAA tournament teams, versus their performance otherwise. These tables suggest, in a very limited sample, that maybe Fazekas is not quite as good as his overall numbers indicate.
Table Four: Prospect’s per-minute Performance Against Tournament Teams
Table Five: Prospect’s per-40 minute Performance Against Tournament Teams
From Erich’s analysis we see evidence that maybe Fazekas is not someone who will excel against front-line talent (then again, maybe in the NBA he will). But even if that isn’t the case, Fazekas – who I think is a free agent this summer – could still be a valuable NBA player. Even if Fazekas can’t produce as a starter (and again, maybe he can), to produce so much off the bench against NBA second-string players is still worth something.
One last note… I should note that even if the college numbers of Fazekas prove to be prophetic, these numbers are not a crystal ball. Remember, these same numbers tell us that Kevin Durant should be a very productive NBA player. And so far, that hasn’t happened (really).
– DJ
Our research on the NBA was summarized HERE.
The Technical Notes at wagesofwins.com provides substantially more information on the published research behind Wins Produced and Win Score
Wins Produced, Win Score, and PAWSmin are also discussed in the following posts:
Simple Models of Player Performance
What Wins Produced Says and What It Does Not Say
Introducing PAWSmin — and a Defense of Box Score Statistics
Finally, A Guide to Evaluating Models contains useful hints on how to interpret and evaluate statistical models.
Frederic
April 9, 2008
I think they have a coaching problem in Dallas. Having played Fazekas 9 minutes and then sacking him was a big mistake. Of course Johnson is only coach, but I think he has a say in this decisions. And Dirks injury has shown that the reserves at his position are not productive. Fazekas obviously would have been. In my opinion developing a young talent is the highest level of coaching. Instead Johnson sends the young talent away and tries to teach the old (highly efficient) dogs new tricks: trying to turn Dirk into Mr. Tripledouble by forcing him to pass more in the beginning of the season or telling Jason Kidd to score more. I think he gave the Mavericks a new edge after he inherited this team from Nelson by teaching them defense. But his effect has worn out and he is to stubborn to correct his mistakes.
Tball
April 9, 2008
I think the last update on Carl Landry had him over .300 WP48. I don’t know if he as dropped since.
dan fitz
April 9, 2008
can he run his monte carlo sim on the NBA?
dan fitz
April 9, 2008
i’m not impressed with avery either. for some of the same reasons mentioned above. he is not developing guys, very rileyish. no numbers to back anything up, but trying to urn dirk into a back to the basket scorer isnt the smartest. maybe it helps. but how about some double screens . he is a spot up/jump shooter
Tom Mandel
April 9, 2008
I’m having some kind of a brain disconnect, apparently, as I can’t quite figure out Nick’s *actual* contribution in his 220 minutes after, now, 78 games. Obviously, he hasn’t won 18 games for the Clippers.
Calculating his Win Score via your published formula produces the number 84. Dividing the resulting total by his minutes gives Nick a .382 productivity per minute. Is my method correct so far? Obviously, that’s an excellent result, one of the best in the league.
How does one get from this number to his real contribution to date?
Jason
April 9, 2008
Tom, “win score” and “wins produced” are not the same thing. The 17.9 refers to his average win-score per 48, not to a number of wins produced.
Michaelangelo
April 10, 2008
Does your analysis of Kevin Durant take into account that he’s been playing out of position all year? The fact that he’s playing as a 2 might have something to do with the lessening of his wow factor.
Tom Mandel
April 10, 2008
Jason — I thought Win Score was a layman’s version of Wins Produced. That is how I read the explanation of it in any case.
Dave, maybe you can take a moment to be helpful here? How does one take the Win Score calculation over 200 plus minutes and extrapolate it to understand what this contribution has meant to the team?
dberri
April 10, 2008
Tom,
Fazekas has a WP48 of 0.316. Therefore in 207 minutes he has produced 1.36 wins (0.316/48 * 207).
You can estimate WP48 from Win Score. To do this you need to calculate Win Score per minute. You then subtract off the per-minute position average. This calculation gives you PAWSmin (or per-minute Position Adjusted Win Score). If you read through the article on PAWSmin, I think there is a simple formula that converts this into WP48.
Of course, that is just the estimate of WP48. In this case, I already have WP48 calculated so we can just work with that.
Jim G
April 10, 2008
I’m curious, in Table One, where does the position adjustment come from for college players?
I’ve never seen position adjustment data for the college level.
David
April 10, 2008
I’m sure Carl Landry has a bigger Wins Produced Per 48 Minutes. Just look at Houston’s records since he started getting regular minutes.
Erich
April 11, 2008
I was quite excited to see Fazekas get some run, but it seems to be done with Brand back. I hope some team gives him a better look next year.
Another favorite from last year’s draft is Lasme. He’s getting some decent time in with the Heat, though his WS/Min looks more like .150 on a quick calculation.
Looks like Win Score again did a decent job of identifying the second round gems, with hits such as Fazekas, Landry, and Lasme.
Erich
April 12, 2008
Dan Fitz,
I can run a Monte Carlo sim on the NBA playoffs, but that would take some retooling.
If I may, I suggest keeping an eye on John Hollinger’s odds at ESPN.com.
I believe Hollinger & Berri agree on the importance of point differential, although they may have their differences when it comes to individual player contributions.
Hollinger has been running a playoff odds report (see link on name) and I expect he’ll run a similar report once the seeds are locked. Given that he’ll be running pretty much the same simulation I would, I’m going to work on other things and keep an eye on his simulations.
Lucky for him, he gets paid to do this stuff!
Enjoy,
Erich
PS. I also suggest watching the odds on tradesports.com as a comparative. There’s a lot to be said on the power of prediction markets…
antonio
April 13, 2008
“Looks like Win Score again did a decent job of identifying the second round gems, with hits such as Fazekas, Landry, and Lasme.”
A little bit early to make this prediction, don’t you think? Especailly for Fazekas and Lasme. Also, I don’t know if there is any truth to this, or if a study has been done, but in my mind it is hard to take Lasme’s stats too seriously. He seems to be a pretty good rebounder, but the fact is with rebounding, no matter how bad the players are, a team will probably still get at least 60% of dreb and 15% of oreb or something like that, so somebody HAS to do that. And the Heat are so miserable right now, with almost every semi-significant player injured, that in my opinion bad players are putting up better stats than they are capable of, simply because somebody has to rebound the ball, score the ball a little, etc.
Erich
April 13, 2008
I look at the second round to generally contain players that do not make valuable contributions to their teams (if they even obtain a roster spot). Given that disposition, I find Win Scores analysis to be an effective means of identifying second round players that do positively contribute to their NBA teams.
As you may agree, the most playing time chances for these types of players seem to be coming from the downtrodden & injured teams. Where we seem to disagree is whether that success would carry over to a winning team. I’ll say yes and unfairly spout: Carlos Boozer. (see link on name, Usage stats, WS/40.)
Cleveland won 17 games in 2002-2003 before LeBron came on board and Boozer is now a key contributor on Utah, posting a similar WS/40 than he posted on a tanking Cleveland team in 2002/3.
antonio
April 16, 2008
throwing out one player really does not prove anything. especially a player who was as good as boozer. boozer was already a borderline star in cleveland. the players you mention have played insignifcant minutes and are not stars. i said especially fazekas and lasme because while they have been productive, it has either way too limited minutes or been producing a task that somebody has to do, regardless how bad the team is (rebounding). I believe if you take the people who have the 12 worst win scores, some of them will be forced to improve because certain stats will be gained regardless.
Erich
April 19, 2008
Antonio,
You are correct that one anecdote does not prove anything, though coming out of the draft as a 2nd rounder, Boozer wasn’t a star. He posted a strong win score in his rookie year which has slightly improved since he’s been recognized as a star player.
My point in picking Boozer was that Win Scores liked him coming out of college, he posted a phenomenal win score in his rookie year on a 17 win team, and he still produces a high win score on a 50+win team. At such a low cost, fans should hope their team takes similar risks in drafting/signing guys like Fazekas. Not all stories will turn out as well as Boozer, but obviously the rewards can be enormous.