Editor’s Note: Shawn Ryan sent some thoughts before the 2010 draft. But time ran short and these thoughts were not posted. Shawn, though, was nice enough to use his analysis to create the post-draft tables. What follows are tables based on his analysis, some of his pre-draft thoughts, and some stuff the editor (DJ) threw into the post. So this is essentially a co-authored post. Thanks again to Shawn for getting this material put together.
Shawn Ryan is a 24 year old student currently living in Austin, TX and pursuing dual degrees in Computer Science and Economics. He found the Wages of Wins at a time when he had become newly infatuated with the field of behavioral economics, and it has greatly influenced his educational goals. He has been a fan of basketball from a young age, but has tended to have different favorite teams over time, including the Suns, Cavaliers, Blazers, Hawks, and Rockets among others. He loathes NBA play-by-play commentary, and often has to resort to turning off the sound for the sake of his mental welfare. He is glad to have the opportunity to contribute to this blog, because for some reason, his fiancee tends to fall asleep at unfortunate times while he discusses his very interesting ideas about the NBA with her.
It’s that time of year again. A time when the NBA’s less fortunate fans, whether they hail from New York, or Minnesota, New Jersey or Memphis, can look to the future and just know that things are going to improve. The NBA draft is a time of infinite possibilities, a time when a single draft pick can change the face of a franchise. Fans hope to nab the next MJ, Lebron, Shaq, or Duncan. Does the 2010 NBA draft hold the next big thing, or just the next big letdown? In the following, I hope to offer some insight into that question.
Win Score and PAWS40
This insight is going to rely upon Win Score. This is a measurement of player performance first noted in The Wages of Wins (and also in Stumbling on Wins). It’s a simplified version of Wins Produced, built around the notion that points, rebounds, steals, field goal attempts, and turnovers have – in absolute terms – the same impact on wins. Blocked shots, assists, free throw attempts, and personal fouls also impact wins, but the impact appears smaller. For simplicity sake, Win Score argues that these factors have ½ the impact of points (and the factors associated with possession of the ball).
Win Score = PTS + REB + STL + ½*BLK + ½*AST – FGA – ½*FTA – TO – ½*PF
As noted in The Wages of Wins (and also in Stumbling on Wins), a player’s Win Score is impacted by position played. Big men tend to get rebounds and not turn the ball over often. Guards tend to do the opposite. Consequently, to compare players across positions, an adjustment has to be made for position played. This adjustment gives us Position Adjusted Win Score (PAWS), a measure that is actually highly correlated with Win Produced.
Before we get to analyzing the NBA draft we need to adjust for minutes played. Specifically, since a college game lasts for 40 minutes, we consider PAWS40. This is calculated as follows:
PAWS40 = Positions Adjusted Win Score per 40 minutes =
(Win Score per 40 minutes – Average Win Score per 40 minutes at position played) +
Average Win Score per 40 minutes for all players
As noted, to calculate PAWS40, one needs to know the average performance at each position. From 1995 to 2009, the players selected out of college at each position posted the following average WS40:
PG: 7.40
SG: 8.40
SF: 9.95
PF: 12.59
C: 12.32
The average player across all position had a WS40 of 10.17. With these numbers in hand, we can calculate each player’s PAWS40. For exampe, point guard John Wall – the number one pick in the draft — had a WS40 of 7.2 in 2009-10. So his PAWS40 would be…
John Wall PAWS40 = (7.2–7.4) + 10.17 = 10.0
A mark of 10.0 is just about average for a player selected out of college from 1995 to 2009.
Since DraftExpress.com is gracious enough to calculate WS40 for almost every NBA prospect outside of high school, we are able to, with relatively little effort, calculate PAWS40 for every player selected out of college in the 2010 draft.
Analyzing the 2010 NBA Draft
Again, an average player drafted from 1995 to 2009 posted a PAWS40 of 10.2. Here is a list of players selected in 2010 who equaled or surpassed the 10.2 mark:
And the below average player are in the following table.
Here are some thoughts on the Draft
- First of all, it must be emphasized (as has been noted in prior discussions of the draft), PAWS40 is correlated with future performance in the NBA. But it is not a crystal ball. So just because a player is above average, that does not mean he will be an above average NBA player. And below average players with respect to PAWS40 might become above average NBA players.
- Given the first point, it may not be the case that Damion James will eventually be the top rookie selected in the draft.
- In fact, given the link between draft position and minutes played, even if James is a very productive player he may not lead this class in Wins Produced. That honor may fall on DeMarcus Cousins and Evan Turner.
- If we look at the below average players, John Wall, Greg Monroe, and Ekpe Udoh top the list. Again, this doesn’t mean fans of the Wizards, Pistons, and Warriors will definitely be disappointed. Still, the PAWS40 evidence is not encouraging.
Perhaps what is said in Stumbling on Wins would help. This is part of one of the end notes in the book:
Comparing college PAWS40 to NBA WP48 revealed that more than 80% of the drafted college players who posted a PAWS40 that was one standard deviation below the mean managed to post a career NBA WP48 that was below the mean (this was true whether you looked at players after three or five seasons). If you look at players with a PAWS40 that was one standard deviation above the mean, though, between 60% to 65% went on to post a career WP48 that was above average. These results suggest that identifying poor NBA performers with college data is easier than identifying outstanding NBA performers. Or in other words, if you play poorly in college, it’s likely that you will play poorly in the NBA. Excelling in college, though, is not a guarantee of future success.
Again, the mean PAWS40 is 10.2. One standard deviation is 2.8. So a player who post a PAWS40 below 7.3 have a very good chance of not being very good players. Posting a mark above 13.0 indicates that above average performance in the NBA is more likely, but not quite a sure thing.
– Shawn Ryan & DJ
Wages of Wins Journal 
Brandon
June 30, 2010
Ok John Wall had a PAWS40 of 10.0 his rookie year which you are saying is average among all players drafted out of college. This basically tells me nothing as all players who are drafted differ in age. The average senior drafted will have a significantly higher PAWS40 then your average freshman. A much more informative final number would be an AGE-ADJUSTED PAWS40.
Arturo
June 30, 2010
Shawn Ryan & DJ,
Cool post. The PAWS40 data is useful but fails somewhat because of the sample. The players are not playing a similar set of opponents and that skews the data. Add age into the equation and it becomes a mishmash.
marparker
June 30, 2010
I don’t necessarily agree with the competition adjustment. These players are spending minimal time on the court against other above average nba talent. I haven’t shown a need to adjust for schedule.
Shawn Ryan
June 30, 2010
Arturo-
Yeah. I was really time crunched on this one, and wasn’t able to go over it subjectively and kind of figure out what was what. I just finished rewriting the post as a Post-Draft analysis, and have submitted it to Dr. Berri.
I don’t know if he will use it or not as it may be redundant, but I put everything into better context. I focus more on the players who deviate most from average, and take age into account in the discussion of many players. If it doesn’t get posted here, then I’ll host it myself somewhere for those who are curious.
Shawn Ryan
June 30, 2010
marparker-
I tend to agree with your assessment. I have the feeling that there is a at least somewhat unwarranted bias toward the larger conferences. It’s an area in which the research really seems to be weak. I think it would be a really interesting, and potentially valuable issue to look into. Though it may be difficult find sufficient data on small conference players.
james McGuiness
June 30, 2010
Your blogs off-brand.
dberri
June 30, 2010
Hopefully I can get the rest of Shawn’s analysis posted soon.
Arturo
June 30, 2010
Shawn Ryan,
Yeah, we’ve been looking a lot at the draft/combine data and it’s kinda like looking at modeling future stock performance from publicly available data (you have to cross your fingers hope all things are equal and no one is lying to their stockholders). Just look at Michael Beasley’s PAWS40.The best thing I can come up with is some interesting thoughts on optimum draft strategy which will come up hopefully in a future post.
dyeyk2000
July 1, 2010
I think strength of schedule and age definitely should be factors that are taken into consideration. Older players tend to be more physically mature but their ceilings tend to be lower also. And there are players who tend to put up huge numbers in smaller and thus weaker conferences.
Hollinger uses SoS in his version of Collegian Ratings. I believe he uses Jeff Sagarin’s SoS, perhaps this can be integrated somehow. (Haven’t really studied the numbers yet, maybe somebody can figure it out). Link below:
http://kiva.net/~jsagarin/sports/cbsend.htm
tywill33
July 1, 2010
I have a question. If you take the average WS40 of drafted players, convert it to WS48, and then compare it to the average WS48 posted last season, the collective decline is 71%. Is that a fair statement? And it appears that perimeter players decline the farthest (63% or somewhere in there) while centers hardly decline at all.
I tried some shortcut analysis and came up with sharper average declines, and I thought my results suggested that centers would see the largest decline in productivity. I’m going to need to walk the cat backwards on those contentions.
I like the way the guy above me is thinking, but it sounds like he (or Hollinger) is introducing unproven assumptions into his model. For instance, has anyone proven that physical maturity limits potential productivity? How would you even measure that? Wouldn’t babies then have the most attractive ceilings? (Someone else came up with that line, its not original, I just don’t remember who to attribute it to)
Great post, btw.
marparker
July 1, 2010
These are rough estimates:
There are roughly 60 major conference basketball teams. Given that each team has 12 or so scholarships thats about 72o players. Dividing by 4 thats about 180 players per class of which roughly 45 will be drafted of which only about half will ever player more than 4 years in the league.
Now, searching for 20 players in a universe of 720 players does it really matter what sample of the other 700 players the top 20 players played against. The top 3% of players are going to face inferior players every night. Should it matter how inferior? In my own work trying to add strength of schedule has never shown any improvement in predicting which players will be good and which will not.
In now way am I suggesting taking Paws40 at face value. I’m saying the adjustments that need to be made have very little to do with competition. All the future nba players in college are playing against inferior competition by definition of being a future nba player.
Shawn Ryan
July 1, 2010
I tend to agree with marparker on this issue. I think that it is a mistake to incorporate SoS into the model without vigorous research on the real affect of SOS.
The difference between the models presented at the Wages of Wins and those presented by John Hollinger is that the former try to dispense with subjective bias. I believe that these metrics should be as completely objective as possible. If we were to incorporate SoS into the model, then we would be systematizing the subjective observation that productive players from low SoS schools are not a good as unproductive players from high SoS schools. I do believe that there is some affect, but whether that effect changes PAWS40 by a coefficient of 1.01 or 20 or just adds to/subtracts from it is unknowable without vigorous study. Alternatively, it may not be the case that SoS has any bearing on the quality of draft prospects. In that case it is folly to incorporate it into your model. If you want models that systematize a subjective approach to analysis, then Hollinger’s work provides that, and many people find his results satisfying. But I would argue that you have to be more careful than that when putting a number to something given that people have a tendency to substitute those numbers for rational analysis when they are available.
Our objective analysis (we also do subjective analysis, and there’s plenty to be had in my upcoming post-draft analysis, there are just no numbers associated with it) does not claim to be all encompassing. It doesn’t list players in the order in which they should be drafted. It presents the production that each player achieve in his particular situation.
One could incorporate SoS, and age, and eye color, and city of origin into ones model today. I would argue though, that until the research is completed, that your probably better off with a completely objective measure of performance as one of your tools, and then intuitively adjusting for those factors. Even so, those intuitive adjustments are likely to be wildly inaccurate as humans also have a tendency to put a higher than warranted weight on the factors about which least is known. To wit, I really believe that you are safer going with an older player who has been productive than assuming that a younger but unproductive player will become productive because he is younger. Better decisions, at least if you are careful and rational in your analysis, cannot come from less information. Think about it. If you can buy them for the same price, i. e. one draft pick, which player do you want to buy stock in, the longer tested player with proven production, the General Electric player, if you will, or do you want the young, volatile, and to this point unproductive player, or the startup player? You only take the startup’s stock if you get it at a massive discount.
To illustrate this point (and now I’m thinking that this argument should have been made in my post, but it’s just come to me, so…), the 29th pick, center Daniel Orton had a WS40 of 8.4 this, his rookie year. Undrafted center, Omar Samhan had a WS40 of 11.2 his freshman year. If you we were limiting ourselves to looking at freshman production, Samhan would be the clear choice.
Samhan’s numbers illustrate another problem with taking picks based on their freshmen numbers. It is widely assumed that player production will boom after freshman year, but Samhan’s WS40 only increased by 0.5 to 11.7. He increased to 14.1 and 15.7 in the following two years. This is certainly not the trajectory that we have in our heads concerning player progression, and it shows that young players are very volatile. Let’s look at another. Zoubek’s freshman year WS40 was only 6.5. The very next year, his WS40 was 12.2. Jr and Sr WS40 were 12.6 and 15.9. His production doubled his sophmore year, but then kind of plataued. Craig Brackins: fre 5.6, soph 10.9, jr 8.7. That sophmore year certainly looked intriguing, but then he fell off. Why does nobody consider that this can happen for young players? To me, this all indicates that youth is not the virtue that we often think it is. And these are just the players that were still NBA prospects by the time that they got to their JR/SR years. I’ve not shown any of the once promising burnouts that almost certainly make up the larger contingent of college Freshman. Now, I would say that if two players have similar production, in a given year, then the younger of the two is likely the better choice, but I think that it is foolish to assume that a Freshman who achieves a WS40 in the 9 or below range is a better pick than a senior with a WS40 in the 12 or above range. Youth, like scoring, is often overvalued in the NBA.
Daniel
July 3, 2010
The problem with Damion James is that he played PF almost exclusively in college, and has limited perimeter skills, though his three-point shooting improved throughout his college career. He’ll probably be playing on the perimeter in the NBA, which is a bit like asking DeJuan Blair to play the SF position (though Blair was a far superior college player).
THAT is the reason he wasn’t picked very high.
Shawn Ryan
July 3, 2010
I’m sorry Daniel, comparing Damion James to DeJuan Blair ridiculous. He plays the wing on offense. He runs his defender all over the court. He cuts to the basket. He runs the court on the fast break. He’s a wing player that rebounds very well, and is not a great shooter or ball handler, but you don’t have to handle the ball much to play SF, and you certainly don’t have to be a good 3 point shooter if you contribute in every other aspect of the game.
In other word, I repudiate your argument completely.
dyeyk2000
July 3, 2010
Shawn Ryan,
Understand your concern on subjectivity but I understand these SoS ratings were derived by using win-loss records and score margins which I would argue represents a more than an acceptable degree of objectivity. This is based on the correlation between good teams and their high average score margins.
On Age, I agree that incorporating some kind of weighting system may be too subjective, but I’d argue that we still need to do this, possibly by correlating it with historical data, and then through a separate valuation system from PAWS40.
After all, as the article suggests, the above numbers would be difficult to decipher unless we take it into context of history (80% of those 1 SD higher go on to be productive players etc.). I’d argue that we can do the same when we incorporate age (ex. 80% of those 1 SD above and 19 yrs old turn out above average etc.)
Shawn Ryan
July 3, 2010
-Dyeyk2000
I do not question the subjectivity of the SoS ratings. My problem with incorporating them is that there is no research on how SoS affects how well a given players stats will translate to the NBA, which ultimately is the important question. The Wages of Wins looked specifically into the matter of how college PAWSmin translates to NBA WP48. There has been no comparable study, to my knowledge for SoS.
Same goes for age.
I do think that there is some affect from at least age, if not SoS, but we just don’t have a study that’s shown that definitively. Until then, I wouldn’t incorporate it into a single metric, because it would be just as likely to mislead than to be helpful.
“On Age, I agree that incorporating some kind of weighting system may be too subjective, but I’d argue that we still need to do this, possibly by correlating it with historical data, and then through a separate valuation system from PAWS40.”
What you’re talking about here is some pretty heavy stuff. If you want to do it right, it will take work. It’s not easy to isolate variables in historical data.
william
July 11, 2010
Cheating is cheating. Whether you get caught or not. That is the problem with sports today. Players will do anything to get an advantage. In the big leagues, that leads to bigger paychecks.
Tom Mandel
July 12, 2010
Shawn — you use “affect” repeatedly when you mean “effect.” Doesn’t invalidate your argument :) but still…
The problem is the unexamined translation from a statistical fact (across a wide range of data, here is a pattern) to individual fact (this or that person is likely to do X, Y or Z.). It’s insufficient to say “of course this statistical fact doesn’t mean that joe blow will be bad”. The statistical fact has literally *no bearing* on whether Joe Blow will be good or bad. How could it?
To take the most obvious example, John Wall did almost everything quite well at Kentucky except that he turned the ball over 4.6 times every 40 minutes. Had he turned the ball over 2.6 times, his WS 40 would be 12 instead of 10.
Now, if one could pull/analyze data on improvement in this stat for kids of 19 as they go forward, wouldn’t that be interesting? Yes, but it still tells you nothing about John Wall (or Joe Blow — were there someone w/ that name). It just can’t. That’s not what statistics is for!
What paws40 certainly *can* tell you is that GMs don’t pick players in the draft with a great deal of success. Of course, we knew this just by looking at their picks! :) But, still, it’s good to have statistical confirmation.
They’d do better if they just picked in order of paws40, no doubt. Better *over all* than they now do *over all.*
Let me put it to you another way — what confidence do you have that Jarvis Varnado will be the 5th most productive player out of the 2010 draft? Or that Dexter Pittman will be more productive than Derrick Favors?
Doesn’t make paws40 useless — far from it! It’s extremely useful. But as with most systems of thought, it’s a slippery slope sliding down quickly to a belief system.
Tom Mandel
July 12, 2010
To put the above in a single sentence: “To a hammer, everything looks like a nail.”
Shawn Ryan
July 12, 2010
-Tom Mandel
Thanks for the comment. I do understand the difference between affect and effect, but it’s one of those grammatical subtleties that that I wasn’t aware of until relatively late (7th grade?), so sometimes, if I’m not careful, the wrong one comes out.
I’m not exactly sure which of my commentary you are referring to in your comments, but I’ll try to address the points that you bring up.
First of all, it seems that something has led you to believe that I put a bit more stock in these numbers than I actually do. Apparently you’ve read my comments about incorporating SoS and other things into draft analysis. Basically I think that it’s all interesting, but shouldn’t be used to make decisions unless it has been studied to a somewhat rigorous degree. I am not however trying to imply that any of these numbers predict how any specific case will play out; as you said, that is just not how statistics work. College PAWSMin (and by logical extension PAWS40, PAWSmin * 40, which is used in the post) has been shown to correlated positively with WP48, which for reasons too complicated to detail here, I feel is the metric which best describes a given players contribution to team wins. Also, PAWSMin and WP48 are quite similar metrics, WP48 just does some more advanced adjustments, and the weights used in PAWSMin are rounded so that they are easier to remember and work with, and thus are somewhat less precise.
So how should these metrics be used? PAWS40 certainly cannot show you whether John Wall or Jarvis Varnado will be a productive player in the NBA. What it can do is tip you off to which players may have a higher than normal probability (after closely examining them for factors that might invalidate their respective PAWS40 rankings) of being productive in the NBA. Really, it probably isn’t wise to choose a player in the draft based on his PAWS40 and not considering any other factors, but it is a good distillation of the production that any given player produced in his given situation, the trick is figuring out how “his given situation” affects the transferability of the production that he achieved. And as of now, there is not a metric that distills the “situation” part of player analysis into a rankable number.
That being said, allow me to respectfully challenge a couple of your characterizations. I don’t like the idea of using statistics simply to confirm our current beliefs, which you allude to, perhaps facetiously. Also, I can’t speak for everybody, but the Wages of Wins is not a belief system for me. This is quite like saying that Darwin is an atheistic scientist’s God. WoW is not a system of beliefs, and I don’t believe that it was designed as such. It is a system of analytical tools which can be used to compare a players box score statistics with those which have been correlated with past wins. There is nothing special about the WoW brand though. There are sports economists other than Berri, it’s just that at this point it seems to me that his tools are the best for this particular niche. Arturo Galletti, who also posts on the wages of wins, is actually working on some original research into player defense that I am very interested in, and which could potentially modify player production rankings offered by the wins produced metric.
As for whether Pittman or Favors will be more productive during their careers, my money is on Favors. He’s much younger and has basically identical production. Plus he will very likely get a lot more minutes which will give him a huge advantage in cumulative production.
Shawn Ryan
July 12, 2010
-Tom Mandel
And to take off on you’re one sentence summary (though I admit, less pithily):
To a psychotic person with a hammer, everything looks like a nail; a rational person can make distinctions however, and uses said hammer only where appropriate.
Tom Mandel
July 13, 2010
“PAWS40 …can …tip you off to which players may have a higher than normal probability (after closely examining them for factors that might invalidate their respective PAWS40 rankings) of being productive in the NBA.”
I couldn’t agree more.
“To a psychotic person with a hammer, everything looks like a nail; a rational person can make distinctions however, and uses said hammer only where appropriate.”
I believe the core of Dave’s work in the 2d book was a contribution to behavioral economics’ challenge to traditional “rational actor” economics. What you call “a rational person” is a model not an entity.
*Of course* one can use a tool properly or improperly. In the right situation or just anywhere. The message of my aphorism was that people often don’t.
Often, because “to a hammer everything looks like a nail” (and for other reasons) they *don’t* make rational decisions. In just the same sense, often a terrific set of ideas — like those in The Wages of Wins and Stumbling on Wins — devolves into a belief system.
Heck, even Dave sometimes uses his ideas that way! E.g. questioning whether JW was the right #1 choice in the draft based on his freshman paws40. Now, this blog is not research — it’s entertainment and journalism — so no big deal.
Finally, please don’t get me wrong. I’m a card-carrying WoW guy! I’d almost certainly have picked deMarcus Cousins first in this draft. But I would have known I was taking a risk and that his freshman on-the-court productivity (via whatever metric) was only one factor in making a decision.
As to “affect” vs. “effect” — please forgive me; I was an English teacher for a while a long time ago, and I can’t help myself. :)
shawnfuryan
July 13, 2010
-Tom Mandel
“What you call ‘a rational person’ is a model not an entity.”
This was probably poor word choice. I wasn’t meaning to evoke the rational person of, for instance, efficient market theory, or that which Richard Thaler might call an Econ. I meant merely to point out that to a person who is careful with the limitations of the tools that he uses, it is possible to avoid seeing any given tool as the ultimate source of all truth.
As you say, this is a blog whose goals include entertainment. And Berri has certainly not shied away from making it known that he thinks that his tools make him capable of judging NBA decisions, but at the same time, he always tempers any proclamations with plenty of caveats. If each post were an academic treatise, then certainly more caveats would be appropriate. Still, I think that Berri strikes a good balance between reminding the reader that rankings are not absolute, that they may change for individual cases, etc. and having 3 pages of endnotes to go with each article. I have tried to emulate this in my writing for the site to a degree.
I actually wrote a much longer and more in-depth post-draft analysis article. It’s been done for a while, but Berri has been busy and hasn’t posted it. Not sure if he will or not, interest in the draft seems quickly to have waned next to the juggernaut of this years free agency period, but you seem to still be interested in the draft so I’ll try to post the article. It’s much more in depth than what you see here, and does put things into context a little more. I’ll leave a link here in the comments once I get everything sorted.
As for judging the Wizards harshly for taking John Wall 1st, I don’t think that that is inappropriate. Though in the article mentioned above, I do kind of give them a pass because they would be seen as incompetent if John Wall somehow became great, and whoever they took instead performed worse than PAWS40 would indicate likely. As a GM, you really can’t afford to take a controversial pick at #1 unless your owner is completely on board.
Finally, there is no need for you to seek forgiveness for providing constructive criticism (although some would argue that grammar-nazi-ing, an activity in which I have partaken on this forum, mind you, is constructive =P , I think it is though), at least not when commenting on my output. I welcome it in fact.
shawnfuryan
July 13, 2010
Here’s the article. Excuse the unfinished aspects of the blog, I just threw this together quickly.
http://shawnfuryan.wordpress.com/2010/07/13/2010_post-draft_analysis/