Today’s guest post is by Erich Doerr. Erich first contacted me prior to the 2006 NBA Draft with a statistical preview in hand. Each subsequent year has seen improvement in the depth and breadth of his analysis, and this article kicks off the WoW Journal’s 2008 NBA Draft coverage. Outside of his basketball writing, Erich does consulting work for major software products by day and has started a fledgling sports-themed Open Source software initiative by night.
It’s May. Where is your favorite basketball team today?
Most likely, they’ve Gone Fishin’ and fans like yourself are left with a lottery ticket and a long off-season. With the 2008 NBA Draft Lottery looming, which of the incoming players should your team target for the franchise’s future?
A Win Score analysis of NBA prospects appears to be an interesting and insightful approach to forecasting NBA success. In the coming week, an analysis of the 2008 NBA draft class will be posted, but first, let’s reflect on a similar assessment of the 2007 draft class.
Scoring on Win Scores
In reviewing last year’s assessments, I will consider the draft slot of the player selected and their ‘07/’08 Win Score, grading on a binary pass/fail basis.
Table One: Review of the 2007 NBA Draft
Win Scores highlighted favorites focused on players not likely to be taken in the first round. The three players identified were Nick Fazekas, Stephane Lasme, and Rashad Jones-Jennings. Fazekas and Lasme were chosen with the 34th and 46th picks respectively while Jones-Jennings went un-drafted.
On the other side, Win Scores seemed rather confident in expecting sub-par play from highly touted lottery picks Corey Brewer, Spencer Hawes, and Acie Law. The following table shows each NCAA to NBA prediction along with a grade.
After reading David Berri’s recent review of the All-Rookie teams, readers may wonder why players like Carl Landry and Rodney Stuckey aren’t listed here. The reason is that coming into the 2007 draft, their performance did not indicate differences from professional scouting rankings, so no prediction was made.
Where predictions were made, Win Scores shot around 70% on NCAA players, which seems to be a lot better than what we see from most GMs.
On the international side, results clearly weren’t as successful. After a heavy dose of assumptions on unavailable statistics, Yi Jainlain, Rudy Fernandez and Jonas Maciulis seemed good enough to endorse while Tiago Splitter warranted pessimism. Yi was the only one to get NBA minutes, though his results were a disappointing .132 WS/M. For the time being, judgment will have to be reserved on the other three.
Quick Thoughts on the Future
How can these results be improved? First, a dose of humility may help in international player assessment. It appears league strength makes international league comparisons difficult for the time being. Second, we’ll be paying much more attention to strength of schedule and collegiate pace in projecting prospects, given some much appreciated reader feedback. Expect these improvements and more in the forthcoming article on 2008 NBA Draft Prospects.
- Erich Doerr
21 responses so far ↓
Mountain // May 17, 2008 at 1:00 pm
I’d suggest , ideally, using strength of schedule defense to adjust offensive stats.
Player defense… I’d add pro prospect points based on scouting or in lieu of that conference awards.
JTapp // May 17, 2008 at 2:34 pm
Since you can calculate PAWS, is it possible to generate Wins Produced for these players? Is there a non-proprietary site where you can get the data necessary to generate PAWS? (I find it hard to find a site that will let you see the attributes that go into WinScores for every player and easily compile them).
porteno // May 17, 2008 at 4:00 pm
always a good read, Erich.
Tom Mandel // May 18, 2008 at 4:29 am
Note that your table, a .png file, doesn’t load in FireFox (2.0 / XP).
I take it that by “successful pick” you mean that your analysis was correct — it’s a little ambiguous, as for a while I took it to mean a successful *draft pick*, and the scores seemed to be reversed in most cases.
Very useful.
Tom Mandel // May 18, 2008 at 4:48 am
Converting Win Score to WP: Dave has provided a formula for an approximate conversion in a comment to another post. It would be useful to have that again, this time somewhere easier to find.
TK // May 18, 2008 at 8:30 am
Tom: That .png file works fine in my FireFox 2.0/XP setup.
FYI,
TK
Tom Mandel // May 18, 2008 at 10:47 am
Converting Win Score to WP:
Dave has provided a formula for an approximate conversion in a comment to another post. It would be useful to have that again, this time somewhere easier to find.
(I restarted Firefox and now the .png works for me too, tk)
dberri // May 18, 2008 at 11:10 am
Tom,
Almost every NBA post (although not this one) has a link to the following column:
Introducing PAWSmin — and a Defense of Box Score Statistics
In that column is the following formula:
WP48 = 0.104 + 1.621*PAWSmin
antonio // May 18, 2008 at 11:57 am
I’ve said this many times, and while I realize there is probably interest in tihs post, while not as bad, it is just like Mel Kiper giving grades a day after the draft. What does it really mean??? In conclusion, not much. The rookie year does not define a player, and judging whether or not winscore or gms or whoever did well is silly this early. While I understand the interest in people looking at the 2007 draft, the earliest time to look at it wuld be after three years, so it will be easier to tell what kind fo palyers these guys will end up being
Erich // May 18, 2008 at 8:17 pm
Mountain,
Don’t forget defensive stats are involved in calculating WS too, so you have to consider opponents defense & offense when evaluating, hence, I’ll be using something like KenPom’s team ratings in assessing strength of opponent.
JTapp, Dave and I discussed using Wins Produced for this analysis, but there are some difficulties involved. I will continue to look into the issue and should have something more to say before the draft.
Antonio, I’d love to grade a draft 3 years after I’ve posted predictions, but unfortunately, I just started publishing these predictions last year. Expect future review articles to cover all previous years predictions, building an online track record as we go. At some point, you may also see a historical review of Win Scores & the draft, and off the top of my head, I believe Win Scores has been very successful in years past.
Oh, and if I had Mel Kiper hair, I’d totally get my picture posted somehow…
Thanks for all the feedback! I’ll try and use a more widespread file format going forward.
Mountain // May 19, 2008 at 7:53 am
Erich,
My suggestion preceded from knowing that rebounds, steals, blocks and fouls represent about half of defense. The other half is shot defense, individual and team. As long as PAWS, PER, NBA efficiency lack shot defense in them directly and reasonably accurately they will be offense biased stats. They are more affected by opponent defense as a whole than opponent offense. Hence my suggestion to adjust the more heavily offense player data by strength of defense.
I shouldn’t have said “ideally” with regard to my suggestion though. There are still compromises.
Ideally a player’s offense would indeed be adjusted by strength of defense faced but also vice versa.
Without good shot defense data for the individual (faced and applied against opponents and for themselves distinctly for a good share of the weight as opposed to team level only for the entire weight) the effort will fall short of ideal. I compromised in suggesting adjustment of player offense by team level opponent defensive efficiency. Most of the focus is on the offense side of the player so that is where I focused the adjustment. But you can do it your way too. Each approach has imbalances / compromises.
Westy // May 19, 2008 at 9:35 am
Wow, I’m looking forward to this.
Dave, thanks for having Erich on to do this. I certainly will be interested to see if Beasley and Rose are clearly set apart like in other rankings.
Mountain // May 19, 2008 at 10:44 am
NCA65 vs other split was suggestive but it would help to see average offensive and defensive efficiencies for each group- for complete set and for the specific set faced by that player.
Mountain // May 19, 2008 at 10:52 am
Using overall strength of schedule helps but variations in the offensive /defensive efficiency mix in schedules / conferences is detail worth knowing, even if it is just on the side and player stats aren’t distinctly adjusted for each side of the court’s action.
Mountain // May 19, 2008 at 11:00 am
I wonder if the stats support that the gap in quality of opponents for college big men is bigger NCAA65 vs all other than for perimeter players and if the gap is bigger college to the pros as well?
The short supply of quality bigs thesis suggest that. If solid then big men translation into pros outside the very best should be tougher?
Perimeter players who rely on outside game might have easier time than college players who rely on going inside if the quality of big man defense is notably tougher in the pros?
VS // May 19, 2008 at 11:09 am
I mean, is it really surprising that when you use the same metric, players play fairly consistently? Your system seems to evaluate players mostly based on how their rebounds and shooting percentage compare to the rest of the league. I don’t see why anyone would really expect that to waver a terrible amount across leagues.
The fact that NBA teams still think nothing of your metric should be a major concern.
Joe // May 19, 2008 at 2:55 pm
As an aside,
Extenuating circumstances are often times missed by statistical analysis. Thaddeus Young is a prime example.
Thad wanted to go pro from High School. He couldn’t due to the age restriction. He decided to go to only a college that was willing to let him write the rules. Georgia Tech said they would only play him at the 3, which Thad demanded. And also gave him the freedom to shoot jump shots all day to showcase his perimeter game for NBA scouts.
When he got to the NBA, he was thrown on the court, primarily at PF, and was told to work the baseline and finish at the rim. As the season progressed, the coaching staff allowed him more freedom to shoot jump shots from time to time and his WS skyrocketed.
This caused the “pleasant surprise.”
Just a little story for anyone interested.
Erich // May 19, 2008 at 5:24 pm
Mountain,
All valid points and plenty of room for further analysis, though there is a bottleneck with available data. Personally, I only have game log data for a select group of players from this year and last, and some more advanced data just isn’t available to the public or just plain doesn’t exist for the college level. Statistical analysis in basketball has a lot of room to grow, and for the time being, I’m giving you my best assessment based on the available data.
(side note: I do like that big man/big conference hypothesis, and may get around to testing it). Thanks a ton for the feedback, as this is how analysis gets better.
Westy,
I think I owe you and Mr. Parker some thanks on your suggestions last year, and I’m glad you are looking forward to the material this year. Enjoy!
VS,
That is a fair critique, though I am writing this article for readers of this blog, a group I would associate as Win Score early adopters. I find WS analysis to be interesting, insightful, and worthy of measurement against other predictors of NBA performance. Over time, competition among approaches brings improvement and weeds out inferior methods, and I’d like to build a prediction track record and compare that against other professionals predictions, be they from the scouting ranks or of the ABPRmetrics aficionados.
Joe, thanks for the aside. I often miss those finer points and appreciate the heads up. Unfortunately, it is difficult to build such information into a model, so we will all be surprised a little bit now & then.
Mountain // May 19, 2008 at 6:34 pm
Thanks for the responses to many and look forward to reading the article.
Erich // May 20, 2008 at 5:32 pm
That lottery combination was the 239th most likely out of 2184 possibilities (0.0813974%).
The 2008 NBA Draft Preview « The Wages of Wins Journal // May 27, 2008 at 9:28 pm
[...] Each subsequent year has seen improvement in the depth and breadth of his analysis. This post continues the WoW Journal’s 2008 NBA Draft coverage. Outside of his basketball writing, Erich does [...]