Each year I like to take a look at the NBA college draft with my custom rating program. The idea is to compare the different college players and predict their NBA success. Here’s what I did: I first take the basic stats and adjust them for: - 40 minute game - Team’s strength of schedule - Team’s win/loss record - Players class (FR, SO, JR, SR) - Players age I then take the adjusted stats and place a weighting factor on them (similar to PERs), with some additional custom modifications. I then take the adjusted stats and compare them against a database of over 1,200 college players that either played in the NBA, or were drafted. The player’s rating is adjusted up or down depending upon the average success of players that had similar statistics. (For example, if the player’s adjusted college stats are very similar to Tim Duncan’s college stats, their score increases.) I then scale the score 0 – 100: 100 is a hall-of-famer 60 is a solid starter 50 is the break between a starter (higher than 50) and bench (lower than 50). 40 is a solid bench player 0 is an absolute bust (played less than a few NBA games) I call the final score the BdB score. This is not a mock draft, but rather a listing of the results of the BdB scores. It only includes those players that have college stats from the previous season. No foreign players are ranked. Here are this year’s results: 1. John Collins, 77.6 Wake Forest (So), 6’10”, 225 19.2 pts, 9.8 reb, 0.5 ast 2. TJ Leaf, 74.8 UCLA (Fr), 6’10”, 220 lbs 16.3 pts, 8.2 reb, 2.4 ast 3. Caleb Swanigan, 71.6 Purdue (So), 6’9”, 247 lbs 18.5 pts, 12.5 reb, 3.0 ast 4. Jonathan Isaac, 67.7 Florida St (Fr), 6’11”, 205 lbs 12.0 pts, 7.8 reb, 1.2 ast 5. De’Aaron Fox, 65.3 Kentucky (Fr), 6’4”, 171 lbs 16.7 pts, 4.0 reb, 4.6 ast 6. Josh Jackson, 64.0 Kansas (Fr), 6’4”, 171 lbs 16.7 pts, 4.0 reb, 4.6 ast 7. Markelle Fultz, 62.8 Washington (Fr), 6’4”, 195 lbs 23.2 pts, 5.7 reb, 5.9 ast 8. Justin Patton, 59.7 Creighton (Fr), 7’0”, 226 lbs 12.9 pts, 6.1 reb, 1.2 ast 9. Eric Mika, 59.4 BYU (So), 6’10”, 230 lbs 20.3 pts, 9.2 reb, 1.6 ast 10. Lonzo Ball, 58.4 UCLA (Fr), 6’6”, 190 lbs 14.6 pts, 6.0 reb, 7.6 ast 11. Dillon Brooks, 57.7 Oregon (So), 6’7”, 215 lbs 16.0 pts, 3.3 reb, 2.7 ast 12. Luke Kennard, 57.7 Duke (So), 6’6”, 202 lbs 19.5 pts, 5.1 reb, 2.5 ast 13. Jayson Tatum, 57.2 Duke (Fr), 6’8”, 204 lbs 16.8 pts, 7.3 reb, 2.1 ast 14. Zach Collins, 56.8 Gonzaga (Fr), 7’0”, 230 lbs 10.0 pts, 5.9 reb, 0.4 ast 15. Josh Hart, 55.0 Villanova (Sr), 6’6”, 204 lbs 18.7 pts, 6.4 reb, 2.9 ast 16. Malik Monk, 54.7 Kentucky (Fr), 6’4”, 197 lbs 19.8 pts, 2.5 reb, 2.3 ast 17. Semi Ojeleye, 51.5 SMU (Jr), 6’7”, 235 lbs 19.0 pts, 6.9 reb, 1.5 ast 18. Bam Adebay, 51.4 Kentucky (Fr), 6’10”, 250 lbs 13.0 pts, 8.0 reb, 0.8 ast 19. Dennis Smith, 50.4 NC St. (Fr), 6’3”, 195 lbs 18.1 pts, 4.6 reb, 6.2 ast 20. Justin Jackson, 50.3 UNC (Jr), 6’8”, 193 lbs 18.4 pts, 4.7 reb, 2.8 ast 21. Tony Bradley, 48.4 UNC (Fr), 6’10”, 248 lbs 6.9 pts, 5.1 reb, 0.6 ast 22. Jonathan Motley, 46.2 Baylor (Jr), 6’9”, 230 lbs 17.2 pts, 9.9 reb, 2.2 ast 23. Jordan Bell, 45.8 Oregon (Jr), 6’9”, 227 lbs 11.0 pts, 8.7 reb, 1.8 ast 24. Kennedy Meeks, 45.1 UNC (Sr), 6’9”, 279 lbs 12.3 pts, 9.4 reb, 1.0 ast 25. Sindarius Thornwell, 44.4 South Carolina (Sr), 6’5”, 214 lbs 21.4 pts, 7.1 reb, 2.8 ast Some others of note: 27. Ivan Rabb, 43.8 28. Og Anunoby, 43.6 30. Donovan Mitchell, 41.3 32. Harry Giles, 39.6 39. Tyler Lydon, 35.1 46. Tyler Dorsey, 31.8 47. Jarrett Allen, 31.0 50. Ike Anigbogu, 28.1 52. Lauri Markkannen, 27.4 John Collins, TJ Leaf, and Caleb Swanigan listed at the top (1-3) is a bit of a surprise. Also, Eric Mika from BYU listed as #9 was not expected. Dillon Brooks is ranked higher than most mock drafts show him going. Lauri Markkannen is the lowest ranked projected lottery player. Of course no rating system or predictive measurement is able to capture all of the factors. This program definitely can’t see 'drug problems', 'injury prone', 'potential', or ‘low basketball IQ’. Misses definitely happen. But it’s fun to give it a shot and see how it turns out in the long run. Here are the links to my posts over the past few years: http://www.sportstwo.com/threads/a-statistical-look-at-the-2015-nba-draft.284772/ http://www.sportstwo.com/threads/the-best-way-to-predict-the-future.215025/ Interesting to see some of those old results. The program grabbed Draymon Green in the top 10, but it missed on Drummond. Each year I adjust the filters to account for the performance of recent drafts. The program continues to grow with more data. (It’s getting smarter!) There are usually around 12-14 starters in most drafts. This program has lined up with that count. This year it's predicting 20 starters. This is definitely a good year to have 3 first round draft picks! My picks for the Blazers this year: If they’re able to trade up – Jonathan Isaac. If they stay where they’re at – (15) TJ Leaf, (20) Bam Adebayo, (26) Caleb Swanigan (One of those big men has got to break out) If they trade down – Dillon Brooks, Jordan Bell (Go Ducks!), or Josh Hart The draft is always a fun time of the year! Let me know your thoughts…
Very interesting.....and a great job! I'd love to see a follow up a year (or two) from now to see how they all pan out. But then, it sounds like you're already doing that. Good stuff! Thanks!
I think it's actually an interesting assessment, I agree it doesn't capture every aspect (in Lonzo Ball's case it doesn't reflect his superior basketball IQ) but it certainly is interesting. What I'd be interested is if you have rankings you've made before previous drafts to see how accurate your predictions are
Yep. I included a few links to a couple of my posts over the years that shows previous drafts. The fun thing about this program is that it grows each year, and depending upon how the most recent drafts turn out, it adjusts it's filters to best line up to improve the results. You would think it's easy in hind-site to write a program that drafts the best players, but when you move one guy up, it adjusts two down. It become a balancing act. Here are the results of previous drafts from this version of the program: 2014: 1. Embiid, 2. J Adams, 3. Smart, 4. K Anderson, 5. Parker, 6. Wiggins, 7. Harris, 8. Gordon, 9. Dinwiddie, 10. Warren 2013: 1. Zeller, 2. Olynik, 3. Porter, 4. Noel, 5. Oladipo, 6. Muscala, 7. Bennett, 8. Dieng, 9. Adams, 10. McCollum 2012: 1. Davis, 2. Crowder, 3. Sullinger, 4. T. Jones, 5. MKG, 6. D. Green, 7. Waiters, 8. Drummond, 9. Lillard, 10. Henson 2011: 1. Irving, 2. Burks, 3. D. Willams, 4. Butler, 5. Marcus Morris, 6. Markieff Morris, 7. K Leonard, 8. Faried, 9. T. Thompson, 10. K. Thompson 2010: 1. Cousins, 2. Monroe, 3. Henry, 4. P. George, 5. Wall, 6. Aldrich, 7. Favors, 8. Whiteside, 9. Booker, 10. E. Davis 2009: 1. Blair, 2. Griffin, 3. Lawson, 4. Curry, 5. Harden, 6. Green, 7. Evans, 8. Carrol, 9. Teague, 10. Holiday 2008: 1. Love, 2. Beasley, 3. R. Anderson, 4. Chalmers, 5. Hibbert, 6. Rose, 7. Douglas-Roberts, 8. B Lopez, 8. Speights, 10. G Hill
Might be interesting if you gave each player's top NBA comparable in your system, if you felt up to it, since you use that as part of your system anyway.
Does your system measure defense? For example on Leaf, here are observations with the eyeball test: "The biggest question marks about Leaf's game translating to the NBA level is where he fits in on the defensive end. He lacks ideal lateral quickness and strength, and doesn't have exceptional length (6'11 wingspan) to compensate. He struggles handling physicality on the block and had a difficult time defending the pick and roll in space, due to his inability to cover ground seamlessly. UCLA often opted to have him drop in pick and roll situations and he was inconsistent with his ability to contain ball handlers getting downhill at the rim. He did average 1.5 blocks per 40 minutes and is a capable rim protector when he uses his leaping ability and reach to go vertical at the rim, but he was often a step slow rotating off the ball and didn't affect a high percentage of shots at the basket. It is hard to say where he will be able impact the game on the defensive end at the next level given his inability to consistently handle bigger players on the interior, or quicker player on the perimeter." - Source: http://www.draftexpress.com/profile/T.J.-Leaf-7242/ ©DraftExpress And does it measure this? "He struggles finishing in traffic, and often avoids contact around the rim as evidenced by his low free throw attempt rate of just 3.8 per-40. -" BTW, I think you have something good here and seems predictive, just not sure about Leaf.
Sure. Keep in mind that part of what this program does is adjust the stats depending upon: SOS, W/L, Minutes, Age and Class. So in some cases it might compare a Freshman player from Creighton to a Senior player from Duke. On the surface they don't have similar stats, but after the adjustments, they came up similar. Also note that this is comparing the college version of each player. For example, Channing Frye did not shoot 3 pointers in college. Here are some of this year’s comparisons: 1. John Collins, 77.6: Raef LaFrentz 2. TJ Leaf, 74.8: Ryan Anderson 3. Caleb Swanigan, 71.6: Zach Randolph 4. Jonathan Isaac, 67.7: Chris Bosh 5. De’Aaron Fox, 65.3: Delonte West 6. Josh Jackson, 64.0: DeMarre Carroll 7. Markelle Fultz, 62.8: Brandon Roy 8. Justin Patton, 59.7: Brandan Wright 9. Eric Mika, 59.4: Troy Murphy 10. Lonzo Ball, 58.4: Michael Carter-Williams 11. Dillon Brooks, 57.7: Gerald Henderson 12. Luke Kennard, 57.7: JJ Redick 13. Jayson Tatum, 57.2: Josh Childress 14. Zach Collins, 56.8: Todd MacCulloch 15. Josh Hart, 55.0: Marcus Thornton (LSU) 16. Malik Monk, 54.7: OJ Mayo 17. Semi Ojeleye, 51.5: Morris Peterson 18. Bam Adebay, 51.4: Channing Frye 19. Dennis Smith, 50.4: Jarrett Jack 20. Justin Jackson, 50.3: Tayshon Prince 21. Tony Bradley, 48.4: Marreese Speights 22. Johnathan Motley, 46.2: Markieff Morris 23. Jordan Bell, 45.8: Patrick Patterson 24. Kennedy Meeks, 45.1: Joey Dorsey 25. Sindarius Thornwell, 44.4: Bonzi Wells 27. Ivan Rabb, 43.8: John Henson 28. Og Anunoby, 43.6: Renaldo Balkman 30. Donovan Mitchell, 41.3: Fred Jones 32. Harry Giles, 39.6: Jarron Collins 39. Tyler Lydon, 35.1: Quinton Ross 46. Tyler Dorsey, 31.8: Wayne Ellington 47. Jarrett Allen, 31.0: Arnett Moultrie 50. Ike Anigbogu, 28.1: Dan Gadzuric 52. Lauri Markkannen, 27.4: Dan Langhi
I agree. I think Leaf is ranked too high, and my guess is that it's exactly what you're noting - defense. Picking up defense is very difficult using only stats. The eyeball test definitely plays a part. The same thing applies to Swanigan. His offensive stats are amazing. 18.5 ppt. 12.5 rpg, 3.0 apg. He's looking like a second rounder primarily because of his defense.
It does take into account where the points come from, yes. Free throws, 2 pointers, and 3 pointers. It also takes into account where the rebounds are coming from. Offensive vs defensive. Good question.
Do you dock PFs for being undersized? Because it's interesting that the top 3 are average to undersized PFs
John Collins 6'10, 218 lbs Wake Forest (So) 19.2 ppg, 55% 2 pt fg, < 0.75 3pt attempted/game, 74.5% ft, 9.8 rpg, 0.5 apg, 1.6 bpg, 0.6 spg Raef LaFrentz 6'11, 240 lbs Kansas (Sr) 19.8 ppg, 62% 2pt fg, <0.75 3pt attempted/game, 73.8% ft, 11.4 rpg, 1.0 apg, 1.5 bpg, 0.9 spg
Yes. That's part of the custom modification I make to the 'PER' formula. Same stat, but taller results in more points.
Be careful with your Kennard to Reddick comparison there @42N8Bounce. Some people don't like that hahaha.
Yeah, I saw the earlier conversation going on. Keep in mind, I didn't make these comps. I input the comp parameters, and the program spits out the player. Luke Kennard 6'5", 180 lbs Duke (So), 19.5 ppg, 5.1 rpg, 2.5 apg, 0.8 spg, 43.8% 3pt, 85.6% ft, 52.7% 2pt JJ Reddick 6'4", 190 lbs Duke (So), 15.9 ppg, 3.1 rpg, 1.6 apg, 0.7 spg, 39.5% 3pt, 95.3% ft, 47.0% 2pt Duke (Sr), 26.8 ppg, 2.0 rpg, 2.0 apg, 1.4 spg, 42.1% 3pt, 86.3% ft, 52.1% 2pt
When your system looks at how the top college comparable performed in the NBA, does it mostly look at prime value or career value? If career value matters, do you account for injuries? For example, you have Brandon Roy as Markelle Fultz's top comp. If Fultz became a Roy-caliber player, that would be a pretty good outcome but I presume he's so low on the list (behind LaFrentz/Anderson/West comp players) because Roy's career value is relatively low. But I assume that doesn't mean the system is expecting Fultz to suffer a career-ending/altering injury. Or do you account for it and Fultz is lower for other reasons?
I like that it rates Fox and Isaac so high; that's especially good for Isaac whom a team would be drafting a lot on potential. And I agree with about where Ball is. Since it doesn't account much for defense it really looks bad for Markkanen to be rated so low.
To measure the value of the NBA career, I use a combination of games played and Win-Share. Players like Roy and Aldridge are a good example of the challenges of ranking players. Roy had a higher peak (Roy 0.115 WS/game, Aldridge 0.109 WS/game), but Aldridge has accumulated more career WS (Aldridge 86.6 WS, Roy 37.4 WS). As for the Roy comparison and where Fultz is ranked, although Roy may have been Fultz's most similar player, it takes an average of multiple players weighted by how similar they are.