Draft Rater: Who will be good pros?
By John Hollinger
Originally Published: June 20, 2011
Many experts think Kyrie Irving is the best prospect in the 2011 draft. What does the Draft Rater say?
It's a fool's errand, but let's do it anyway.
Yep, it's time for the Draft Rater. There are inherent limitations in trying to pore through a player's statistics and project what he'll be capable of five years down the road. The things the stats don't tell us -- about his dedication, eating habits, off-court life, the system his team runs, and 10,000 other things -- badly outnumber what the stats do tell us.
Yet, surprisingly, the stats seem able to tell us quite a bit. That's the premise behind the Draft Rater, my annual data-driven guide to the draft, and this year we have better clues than ever as to how it can help us and how it can't.
To review, my Draft Rater is a regression analysis comparing 16 variables to a player's NBA player efficiency rating, using the average of their top three seasons in their first seven years as a pro. Some haven't played three seasons yet or won't ever, so we take their career PER. We've also set a PER floor of 4.0 for those who couldn't make the league, and 5.0 for those who barely made it.
I've once again rebuilt it from the bottom up this year. Along with this year's revisions, the Rater was already getting smarter every year as it got more data and more NBA results from players already drafted. As a result, we can see with greater specificity which statistics translate to the pro game and which ones don't.
Second, we've seen the particular ways in which it fails. The most obvious one is on all the squishy stuff -- character, dedication, conditioning, etc. Michael Beasley, Michael Sweetney and DeMarcus Cousins all got huge marks from the Draft Rater, but one could justify passing on them on draft day given the other red flags. Similarly, we don't have a good measure for injury-proneness either -- Curtis Borchardt, Brandan Wright and Greg Oden, take a bow.
But more particularly, in back testing this year's Draft Rater, it's become obvious where it succeeds and where it falters. To wit:
• One-and-done gives it trouble. This isn't a fatal weakness, but players who stay only one year don't give the Rater enough information to develop a reliable estimate ... especially ones who improve rapidly through their freshman season, as Derrick Rose did in his one season at Memphis. Rose still finished with a strong rating, but if we'd based it solely on the second half of his freshman season it would have been much higher.
• UCLA messes it up. For some reason, every Ben Howland product massively outperformed his estimate as a pro. This is over a period of seven years involving 13 NBA prospects, and all of them except Josh Shipp outperformed; many of them did so by wide margins. Given the consistency of the disparity, I included a "Howland" variable in this year's model. One can argue that this is a case overfitting the model to past results, so we'll see how it projects this year with Tyler Honeycutt and Malcolm Lee.
• It's way better with perimeter players. This is probably the biggest realization from looking at back data on the draft rater. Perimeter players with high ratings all become stars. All of them. The big guys? Not so much. Some of them have become superstars -- Love, Blake Griffin, Chris Bosh. Many others, however, have merely been decent players. A couple with fairly high ratings haven't been able to play at all.
I believe there are two reasons for this.
The first is off the court. Looking back, if I could tag the stats with additional variables like "doesn't like basketball," "space cadet" or "pothead," I could make the projection for big men considerably more accurate.
This isn't a factor for perimeter players because it's very difficult to have such glaring red flags and succeed as a wing or a point guard -- there are too many other people of the same size and too many skills required that can be developed with only a certain level of dedication.
For big men it's a different game, however, especially at the college level. The biggest and most athletic ones can dominate the glass and shoot 65 percent on dunks and layups without developing much of anything else in the way of basketball skills.
Second, I believe the stats translate better for perimeter players because it's very hard to get true one-on-one post-up chances at the college level, but this is a major part of the diet for most NBA big men -- especially the ones that are drafted in the lottery. So for perimeter players, the college stats are a much more apples-to-apples translation.
• It skews higher for big men. This, in retrospect, is not a "problem" as much as something to keep in mind. In any draft, most of the early picks are big men, and there's a good reason for this -- first because size is rare, and second because bigs in general have a greater disparity between good and bad.
Second, historically power forwards have had the easiest time racking up a solid PER, and most of the big men on the board on draft day are power forwards rather than centers. As a result, in any given season there will be more highly rated bigs than perimeter players.
So, to review, it ain't perfect. But we can still learn a ton from it. Let's take a look at this year's Rater and see what it says about the top college prospects:
Kyrie Irving is the one sure thing
If I had to put my money on one player in the draft becoming a star, it would be Kyrie Irving.
OK, no shock there.
But here's the logic -- Irving's rating of 15.12 is the best of any perimeter player. While he compiled that in just 312 minutes, his low minute total actually hurts him in the Draft Rater (minutes played are a positive indicator of future success). And he's in pretty exclusive company. The other perimeter players to post a rating of more than 15 all became stars -- since 2002, the six perimeter players to do so are Chris Paul, Carmelo Anthony, Kevin Durant, Dwyane Wade, Danny Granger and Rudy Gay. Based on those comparables, I like my chances if I'm Cleveland.
Two more perimeter players to like
If I had to peg two other perimeter players that I would guarantee to at least become solid rotation players, it would be Kemba Walker and Kawhi Leonard. While this year's draft doesn't project to have a lot of star talent at the perimeter positions, Walker and Leonard are the two who rate above 12 -- which, historically, has been a guarantee of at least being decent.
Of the 13 point guards to rate above 12, the list includes Paul, Jordan Farmar, T.J. Ford, Jay Williams, Russell Westbrook, Mike Conley, Jameer Nelson, Derrick Rose, John Wall, Frank Williams, Ty Lawson, Ray Felton and Darren Collison. Jay Williams had a bad injury, of course, but of those 12 only Frank Williams couldn't play. The others were all decent-to-great, which means Walker, at 12.75, looks fairly bust-proof.
Similarly, of the 14 wings to rate better than 12, the list includes Durant, Wade, Anthony, Granger, Gay, Luol Deng, Josh Childress, Mike Dunleavy, Dajuan Wagner, Rashad McCants, Evan Turner, Delonte West, Caron Butler and Ben Gordon. Only Wagner and McCants failed, and each pretty clearly had NBA talent. So Leonard, at 13.21 with no injury or character red flags, looks like a very, very safe play.
The mystery man
Tyler Honeycutt of UCLA is the player I'm most interested in watching in this draft (and this coming season). His rating of 12.56 rates him as a quality rotation player on the wing and a lottery pick.
However, much of his rating stems from the fact that he's a Howland product; eliminating that fact from the database knocked him down several pegs. Again, if we're guilty of overfitting the model to past results, he shouldn't be this high. He's projected to go late in the first round right now and could represent good value based on the overachieving history of other players from his school. Since I'm still apprehensive about the UCLA adjustment I have him 13th on my board.
Two Euros to watch
My Euroleague translations say two players in this year's draft, Jonas Valanciunas and Nikola Mirotic, would be rotation players immediately if they came to the States. Valanciunas rated higher, with a translated PER of 14.70 (albeit in limited minutes) compare to 13.66 for Mirotic. Of perhaps more importance is that Mirotic is contractually bound to stay in Europe for a few more years. He'd be a top-10 pick on my board if it weren't for that; as it is, I've dropped him behind all the players I consider relatively safe bets.
The other top European prospect, Czech forward Jan Vesely, doesn't grade out nearly as well. His first-year PER translates to 10.72; while one supposes he would improve further from that point given his youth and athleticism, it still makes him somewhere south of a sure thing. I've listed him as a top-20 pick based on potential and the general weakness of this draft, but taking him in the top five or 10 would be a mistake.
Two other internationals who warrant mentioning are Giorgi Shermadini and Bojan Bogdanovic. Shermadini, a 7-footer from the Republic of Georgia, forecasts as a sleeper with a 14.63 translated PER; however, he played in just 249 minutes, so we should take that with a grain of salt. He's an intriguing second-round play nonetheless. Bogdanovic is a more traditional second-round hopeful -- probably not good enough to play in the NBA now, but maybe he improves on somebody else's dime over the next few years.
No numbers here
There are four international men of mystery in this draft who did not play in the Euroleague last season and thus have no translated stats for me to discuss: Bismack Biyombo, Enes Kanter, Donatas Motiejunas, and Davis Bertans.
Fortunately, I've seen all three at the Hoop Summit the past two seasons. Biyombo has scouts worried because he can't shoot at all, but he's a dominating defensive force in the paint; at the absolute worst he's going to be better than Ekpe Udoh. I slotted him 11th on my board, behind all the players the Draft Rater is really gung-ho about. He's going to be a rotation player based on defense and rebounding alone; the question is if he can finish enough plays at the basket to start.
Kanter is sort of the anti-Biyombo; He's not much of an athlete and will be suspect at the defensive end, but has such a high skill level offensively that he's going to score relatively easily. A good comparable might be fellow Turk Mehmet Okur, except Kanter is probably more skilled with the ball.
Motiejunas and Bertans are worthwhile choices later in the first round; each is high on skill but suspect in terms of strength and athleticism. An American who was last seen in Europe, Jeremy Tyler, falls into the same category.
Two point guards on the fence
Brandon Knight could go as high as the third pick, while Jimmer Fredette is also a likely lottery selection. Draft Rater is pretty lukewarm on both of them. Knight rates at 10.02 and Fredette rates at 10.45; unheralded Norris Cole of Cleveland State has a better rating than both. Each projects to have a career, but probably as a third guard or marginal starter.
In Knight's case, as a one-and-done we have to acknowledge that the system hasn't rated players like him as accurately, although it has done very well with guards as a whole. Fredette has a slightly better rating in a larger body of work.
Among point guards to rate between 10 and 11, the historical comps aren't great: One All-Star (Rajon Rondo), one really good player (Kyle Lowry), several halfway decent players (Luke Ridnour, D.J. Augustin, Jerryd Bayless, Mario Chalmers), and some end-of-bench filler (Darius Washington, Marcus Banks).
I moved Knight up to 12th on my board to reflect that his one-and-done status may result in his being undervalued; on sheer rating he'd be in the 20s. Fredette I've left at 19 ... two spots behind Cole.
Guards who don't make the cut
Probably the most suspect candidate, according to Draft Rater, is Marshon Brooks; his 7.88 rating was 27th among perimeter players. The Providence guard put up huge stats, but his average is hugely padded by two factors: First, the Friars played the fastest pace of any major Division I team; second, Brooks played nearly every minute of every game, averaging 36.5 per game -- remember, they play only 40 in college. Let all the air out and his numbers look a lot more ordinary -- his usage rate, which is his most alluring stat, ranks only ninth among prospects. Given his age (22 and five months) and his relative inefficiency, I'm not sure there's a ton to see here. Draft Rater sees him as a second rounder.
Kansas' Josh Selby rated even worse -- 7.69, 30th among perimeter players. I moved him up my board a bit to account for his one-and-done status, but his raw numbers were rather poor. In particular, a point guard with a -1.11 pure point rating should send talent evaluators shrieking. It was easily the worst of any point guard prospect, and worse than all but five wings as well.
The big man conundrum
As I noted above, the Draft Rater has been really solid on perimeter players. On interior players, the results have been a bit more scattered. The problem has been "false positives." It has picked out all the guys who could play; it has just picked out a lot of other guys a long with them.
For that reason, we want to tread a little more carefully with the frontcourt players. However, two players in particular warrant our attention: Tristan Thompson and Derrick Williams.
Thompson and Williams had the highest ratings of any player in the Draft Rater this year, and while that doesn't come with the same assurances it does for Kyrie Irving, they both appear to be very solid prospects. Of the 13 players who rated at 15.5 or above in previous iterations, most were very successful as pros, and the ones that weren't tended to fail due to injuries and lack of professionalism -- issues that shouldn't be factors for Thompson and Williams. The one true miss was Jared Jeffries.
The other strong frontcourt prospect is Tobias Harris of Tennessee with a rating of 14.83. Of the 19 big men to rate between 13.5 and 15.5, a few were dogs, but two became All-Stars (Al Horford and Carlos Boozer) and most became quality players.
Things start getting more iffy at the next level, where we get into the Jon Leuers and Nikola Vucevicses. Also included in that group is unheralded Greg Smith from Fresno State, who could end up as a second-round steal.
Big men who rated between 12 and 14, as those three did, have been a mixed bag: Nearly all had careers and some were very good, but several were career backups and a few were just flat-out bad. Of the 29, two became All-Stars (LaMarcus Aldridge and David West), and 15 of the 29 became top-eight rotation players.
Similarly, those who rated between 11 and 12 -- as Malcolm Thomas, Jordan Williams, Rick Jackson and JaJuan Johnson do -- were very much a mixed bag. The takeaway here is to put more credence on scouting reports with players in this range; I've done that on my draft board. Lacking strong evidence one way or the other in this range, we'll go with the subjective opinions.
At the back end, two players who rate surprisingly weak are Marcus and Markieff Morris of Kansas. Historically, big men who rate between 10 and 11 are career backups. Of the 23 who did so, only David Lee became a star; five others became top-eight rotation players, and the rest were bench filler or didn't make the league at all. Based on that information, they're late first- or early second-round selections. On my board, I have the Morrises ahead of the other players in the 10-12 range, but it's tough to justify putting them ahead of similarly rated perimeter players.
Summing it up
Which takes us to the final step -- my draft board. Based on all the information from Draft Rater, the projections of the Europeans, what I've seen in the past three Hoop Summits, and using the general consensus of draftniks as a tie-breaker with the close calls, here's how my board of the top 60 looks.
But first, if you're looking for the CliffsNotes version of my Draft Rater, here are the five big takeaways:
• Marshon Brooks, Josh Selby and the two Morrises appear overvalued.
• Brandon Knight, Jimmer Fredette and Jan Vesely are overvalued, but not as strongly.
• Kyrie Irving and Derrick Williams are, rightly, the top two players on the board.
• Tristan Thompson and Tobias Harris are undervalued.
• Jon Leuer, Norris Cole and Greg Smith are your sleepers.
And now, my top 60 heading into Wednesday:
My board
1. Kyrie Irving
2. Derrick Williams
3. Tristan Thompson
4. Jonas Valuncianas
5. Kawhi Leonard
6. Enes Kanter
7. Kemba Walker
8. Tobias Harris
9. Alec Burks
10. Jordan Hamilton
11. Bismack Biyombo
12. Brandon Knight
13. Tyler Honeycutt
14. Jon Leuer
15. Nikola Vucevic
16. Chris Singleton
17. Jan Vesely
18. Klay Thompson
19. Norris Cole
20. Iman Shumpert
21. Nikola Mirotic
22. Jimmer Fredette
23. Donatas Motiejunas
24. Greg Smith
25. Marcus Morris
26. JaJuan Johnson
27. Markieff Morris
28. Davis Bertans
29. Kenneth Faried
30. Jeremy Tyler
31. Jordan Williams
32. Bojan Bogdanovic
33. Darius Morris
34. Giorgi Shermadini
35. Reggie Jackson
36. Rick Jackson
37. Brad Wanamaker
38. Damian Saunders
39. Nolan Smith
40. Malcolm Thomas
41. Travis Leslie
42. Trey Thompkins
43. Malcolm Lee
44. Charles Jenkins
45. Cory Joseph
46. E'Twaun Moore
47. Josh Selby
48. Jacob Pullen
49. Justin Harper
50. Jimmy Butler
51. Matt Howard
52. Jamie Skeen
53. Jereme Richmond
54. Keith Benson
55. D.J. Kennedy
56. Marshon Brooks
57. Isaiah Thomas
58. Andrew Goudelouck
59. Lavoy Allen
60. Shelvin Mack
http://insider.espn.go.com/nba/draft...ftrater-110620