Okay as per suggested by one of our friends, I will also attempt to rank NBA players (mainly paying more attention to our rockets 5 key players i.e. Howard, Harden, Lin, Asik and Parsons) - but I will need to track the other starters of the other teams since I will be trying to compute some sort of head-to-head comparisons so as to ascertain the changes in their rank values. Again this is something similar to the Chess Elo ranking (with some tweaks thrown in). Some background - I will again stick with a working range of 82 (bottom value - this is fixed) and an arbitary ceiling of 2460 (for starters but this is not fixed for players unlike the system I am employing for the team rankings). For baselining - I will be basing on basic stats provided for players from NBA stats. For new draft players and players without stats I will make use of a provisional value which is usually the average (for now it is 1271). For guards my initial computation for baselining is based on 3 elements - PPG, APG and SPG - I have used equal weights on them (yes, some of you may dispute whether we should tweak with different weights I am opting for something simplistic - you are welcome to try to tweak it at your own leisure). For forwards my initial computation for baselining is based on 3 elements - PPG, RPG and BPG. I haven't got a chance to work on the Forwards stats as yet - but here's what I have calculated as baselines for our guards (note I have estimated for 60 of the starting guards in NBA (but that's on the depth chart presented on some other site (it may change in time). Won't show all 60 just until both our players turn up on the list. For those interested, when the games start in October this rank scores will change as our players perform against different opponents (they may climb or decline depending on their individual performances (i.e. not tied to team performance) in comparison to the average ranks of their opponents. I will give an illustration later. But here's how our guards ranks out based on last season's stats: Rank Score Name 1 2046 C. Paul Clippers 2 1882 R. Rondo Celtics 3 1834 R. Westbrook Thunder 4 1775 J. Harden Rockets 5 1720 S. Curry Warriors 6 1704 M. Ellis Mavericks 7 1688 K. Bryant Lakers 8 1676 R. Rubio Timberwolves 9 1669 J. Holiday Pelicans 10 1619 D. Wade Heat 11 1618 M. Conley Grizzlies 12 1606 K. Walker Bobcats 13 1593 K. Irving Cavaliers 14 1561 J. Wall Wizards 15 1547 D. Rose Bulls 16 1538 B. Jennings Pistons 17 1536 G. Dragic Suns 18 1517 T. Lawson Nuggets 19 1485 J. Teague Hawks 20 1483 D. Williams Nets 21 1449 T. Parker Spurs 22 1440 J. Nelson Magic 23 1400 P. George Pacers 24 1392 J. Lin Rockets I don't think I want to clutter the rest of the details - but as an illustration here's how I would be proceeding to calculate the future changes of values for our players. The first game that Rockets will be playing will be against the Bobcats. The 2 guards for Bobcats are currently carrying the values of K. Walker (1606) and G. Henderson (964) - so the average opponent strength is about 1285. So both Harden (1775) and Lin (1392) are expected to outperform Bobcats guards overall. By out peforming, I mean for the 3 statistical areas of interest Points, Assists and Steals, I am expecting them to score at least 2 to 1 when they play against the Bobcats. If they score more than the averaged score of the 2 guards they would get 1 point, if they draw they would get 0.5 points (likewise for Assists and Steals) - in theory they could score from 0 to 3 poins. But for simplicity's sake I will treat 2 and above as them having performed to expectations. Specifically, I will be expecting Harden to have a 58% chance of outperfoming the rivals and Lin to have 52% - this has an impact in score differentials of gaining or losing points. The points at stake for Harden will be 34/48. If he wins he gains 34 points but if doesn't makes it he loses 48 points. Whilst for Lin, the rate is 39/43 - wins 39 points loses 43 points. If some of you wonder why the "unfairness" - in fact the system is fair in the sense that one would expect a better player to have a better performance so they should be penalized more if they underperform. As mentioned in the other thread - this method is a self-correcting ranking system - i.e. the more we use it, the higher our confidence of assessing the relative ranking score of our players against the opponents that they have played against. For those who may want more details - maybe we can follow up away from the threads so as not to bore too many of our readers. Frequency wise - I am not so sure if I will be up to it - to update as soon as the games are completed - my main priority would remain tracking of the overall team performance - so am likely to see if I can squeeze time somehow to do this. It's not easy though - so if I am not regularly updating this please understand. Feedback and suggestions are welcome (although I can't promise I can always act on them)..... Go rockets!
Ellis at 6 and Parker at 21?? When I get results like this for something I'm working on at work - I scrap it and start all over.
Why should a guy's backcourt teammate influence his rating so dramatically? Say Lin is going head to head against Jennings. Jennings has to beat the average of Lin and Harden. Lin has the much easier task of beating the average of Jennings and Stuckey or whatever, and if things go to average Lin moves up on Jennings even if Jennings beats him pretty handily head to head. Harden just pulled Lin to a victory over a better point guard.
Good question CDrex - my answer is that in a real game - it's not always a head-to-head confrontation, it is more likely that a guard will go against another guard some of the time. (i.e. Jennings isn't exactly going to stick like glue to Lin, he'd have to guard Harden as well at times). Of course, we can debate on what relative weights (or portion of the time mismatching occurs) but for simplicity's sake (and starting) we will go with a simple average and see how things go. The other reason I am going with an averaged opponent score - is that there will be times players are switched and I wouldn't have the baselines for the newly switched in players (it is much better to mitigate any potential errors swings using an averaged score versus using the individual opponent scores). For new players introduced in opposing teams that are unrated I am likely to just use an averaged player ranking score of around 1271 as starters. I think there was another poster asking why Parker was ranked so low? Beats me - I may have to go back and look at the base stats provided by NBA - it might be screwed up (if not the basis is really on one of the factors - since I also have given the same weight to Steals - that isn't really one of Parker's glorious achievements last year). Anyways, we shouldn't take this starting value as something really important - it's just my best guestimate at the moment before the season starts - when the season gets going the value will self-correct itself as Parker plays better his value will quickly raise closer to his actual standards - yep - please note this is not a static value system it changes after everg games the players had played. PS: I have made an error calculating the values (I had forgotten my baseline wasn't zero but 82, so I would need to add an adjustment value to the scores - there isn't any major changes in rankings (just movement of 1 position for some of the players - i.e. J Lin gets bumped up to 23.
You make a good point and I realize this isn't the issue here, but Jennings isn't better than Lin and if he is, it's only incredibly marginally. If you break it down they're pretty much the same guy. ...Monta at 6 though, wow. I guess he does have it all.
I'm really bad at understanding and interpreting advanced stats, so I won't comment on the validity or anything. But, if there's any truth to this: 1) The Rockets (probably) have the most balanced starting backcourt in the league (with an average rank of 13). I wouldn't say this is too far off, as most elite guards have poor running rates (Chris Paul, Rondo, Wade), although I'd maybe put us behind the Warriors. 2) The point guard position is much deeper in the NBA than the shooting guard position (although this seems to be widely accepted). 3) Backcourt talent is rather evenly distributed throughout the league. Only the Rockets have two players in the top 24. Again, this would reaffirm another common held belief, that while quality guards are very nice to have, wingman and big man are more valuable and impactful when it comes to championships (I'm inclined to agree, although I still think there's something to be said for a high powered backcourt). It be interesting to rank all the players, and see which team has the best starting five.
Okay not sure how many of you would like this but here's how I have provisionally ranked the 90 starters (presumed because the starting position may change due to trades, injuries, and coaches liking/disliking and other "basketball" reasons....) Please note this is just provisional ranking (don't worry if your favorite player seem to be ranked too low (e.g. LeBron) - as the season progresses, the system will correct itself because such players scores will rise up in accordance to their performance - it will just take a little bit of time - my own estimation after around 29 games - the rank values should be getting very close to most people expectations - and I hope it would pass the so-called "eye-ball" tests of some of our posters (wish me luck!) - so here goes: Forwards (important elements considered : PPG, RPG, BPG - equal weights) 1. T. Duncan Spurs 2257 2. D. Howard Rockets 2256 3. S. Ibaka Thunder 2114 4. A. Bynum Cavs 2071 5. L. Sanders Bucks 2032 6. R. Hibbert Pacers 1948 7. J. Noah Bulls 1906 8. K. Durant Thunder 1898 8. B. Lopez Nets 1898 10. J. Smith Pistons 1800 11. L. Aldridge Blazers 1713 12. L. James Heat 1704 13. A. Davis Pelicans 1672 14. M. Gasol Grizzlies 1627 15. K. Love Timberwolves 1623 16. A. Horford Hawks 1622 17. A. Jefferson Bobcats 1578 18. N. Vucevic Magic 1547 19. C. Anthony Knicks 1540 20. M. Gortat Suns 1536 21. O. Asik Rockers 1485 22. P. Gasol Lakers 1460 23. D. Favors Jazz 1445 24. T. Chandler Knicks 1438 25. D. Cousins Kings 1437 26. J. McGee Nuggets 1427 27. D. West Pacers 1393 28. D. Lee Warriors 1392 29. N. Pekovic Timberwolves 1390 30. A. Drummond Pistons 1388 31. G. MOnroe Pistons 1386 32. S. Hawes 76ers 1380 33. R. Lopez Blazers 1379 34. A. Bogut Warriors 1370 35. A. Johnson Raptors 1366 36. K. Faried Nuggets 1347 37. Z. Randolph GRizzlies 1338 38. B. Griffin Clippers 1334 39. K. Garnett Nets 1329 40. C. Bosh Heat 1327 41. T. Thompson Cavs 1325 42. P. Millsap Hawks 1323 43. D. Jordan Clippers 1313 44. C. Boozer Bulls 1284 45. K. Olynyk Celtics 1271* 45. O. Porter Wizards 1271* 47. E. Okafor Wizards 1270 47. N. Batum Blazers 1270 49. D. Nowitzki Mavs 1269 50. R. Gay Raptors 1257 51. T. Young 76ers 1232 52. J. Valanciunas Raptors 1209 53. G. Davis Magic 1184 54. S. Marion Mavs 1167 55. P. Pierce Nets 1160 56. L. Deng Bulls 1096 57. E. Ilyasova Bucks 1081 58. N. Hilaro Wizards 1080 59. T. Splitter Spurs 1074 60. J. Thompson Kings 1069 61. D. Gallinari Nuggets 1065 62. S. Dalembert Mavs 1056 63. A. Iguodala Warriors 1054 64. M.K-Gilchrist Bobcats 1040 65. C. Kaman Lakers 1035 66. T. Harris Magic 1028 67. K. Leonard Spurs 1020 68. C. Parsons Rockets 1010 69. J. Green Celtics 1009 70. G. Hayward Jazz 1004 71. C. Zeller Bobcats 1001 72. K. Perkins Thunder 986 73. B. Bass Celtics 958 74. A. Bargnani Knicks 955 75. T. Evans Pelicans 950 76. M. Morris Suns 920 77. E. Turner 76ers 919 78. J. Smith Pelicans 893 79. E. Clark Cavs 892 80. K. Korver Hawks 837 81. J. Henson Bucks 808 82. T. Prince Grizzlies 765 83. E. Kanter Jazz 742 84. C. Budinger Timberwolves 661 85. J. Dudley Clippers 637 86. L.Mbah a Moute Kings 612 87. N. Young Lakers 607 88. C. Butler Suns 600 89. U. Haslem Heat 583 90. D. Granger Pacers 427 Guards (Important elements considered: PPG, APG and SPG) 1. C. Paul Clippers 2060 2. R. Rondo Celtics 1901 3. R. Westbrook Thunder 1855 4. J. Harden Rockets 1797 5. S. Curry Warriors 1744 6. M. Ellis Mavs 1729 7. K. Bryant Lakers 1726 8. R. Rubrio Timberwolves 1702 9. J. Holiday Pelicans 1695 10. D. Wade Heat 1647 10. M. Conley Grizzlies 1647 12. K. Walker Bobcats 1634 13. J. Wall Wizards 1591 14. D. Rose Bulls 1578 15. B. Jennings Pistons 1569 16. G. Dragic Suns 1567 17. T. Lawson Nuggets 1549 18. J. Teague Hawks 1517 19. D. Williams Nets 1516 20. T. Parker Spurs 1482 21. J. Nelson Magic 1474 22. P. George Pacers 1435 23. J. Lin Rockets 1428 24. G. Vasquez Kings 1426 25. D. Lillard Blazers 1390 26. K. Lowry Raptors 1342 27. R. Felton Knicks 1330 28. M. C. Williams 76ers 1271* 28. B. McLemore Kings 1271* 28. T. Burke Jazz 1271* 31. J. Calderon Mavs 1204 32. L.J. Mayo Bucks 1185 33. G. Hill Pacers 1180 34. K. Irving Cavs 1154 35. S. Nash Lakers 1146 36. E. Gordon Pelicans 1142 37. L. Williams Hawks 1089 38. W. Matthews Blazers 1081 39. M. Chalmers Heat 1060 40. D. Derozan Raptors 1044 41. D. Waiters Cavs 1025 42. J. Johnson Nets 1015 43. G. Henderson Bobcats 1014 44. K. Thompson Warriors 1011 45. L. Ridnour Bucks 1005 46. E. Bledsoe Suns 994 47. A. Affalalo Magic 965 48. B. Beal Wizards 925 49. JJ Redick Clippers 918 50. A. Bradley Celtics 900 51. T. Allen Grizzlies 882 52. D. Green Spurs 879 53. J. Richardson 76ers 855 54. K. Martin Timberwolves 847 55. T. Sefolosha Thunder 810 56. J. Butler Bulls 735 57. I. Shumpert Knicks 711 58. C. Billups Pistons 638 59. A. Burks Jazz 536 60. E. Fournier Nuggets 475
I would say there is a major flaw in dividing players into Guards or Forward/Centers, and it is made apparent by the SF position which can play like a wing player OR a big man. Not to mention the point forward position. The fact that there can be point forwards in the league, but APG is not counted to their ranking, makes things interesting. Lebron's low ranking is obviously due to the fact that his APG is not included as part of the analysis. Also seeing players like Ibaka, Hibbert and especially Sanders so high makes me question whether there is truly equal weighting given to PPG, RPG and BPG. By those 3 players, it looks like BPG is weighted much higher. If you are going to really simplify things, then you should remove the Guard vs Forward categories and just make all players the same. You can still put each position head-to-head against each other, but let all positions include PPG, RPG, APG, BPD and SPG. There must be value in a guard being able to put up block against another guard, or for a SF to make plays and have higher APG like Lebron.
Good points dakeem, will take a look at what turns up if we lump all 150 players into one pool and give equal weights to PPG, APG, SPG, RPG and BPG. I also think I am most likely needing to tweak the weights for BPG (as the figure there is generally a very small range so - the swing in value seem to be high when mulitplied out and - okay maybe not the weights when I averaged the values but how I derive a RRS value from each of the elements - i.e. maybe I will need to normalized them before I can put them together. So will think about what normalization process I should use... thanks!
Okay, as per comments and feedback received. I have gone back to the drawing boards to recalculate the Relative Ranking Scores - what have I changed? I took a look at all 5 elements PPG, APG, SPG, RPG and BPG. Then I tried to normalized the values by adjusting values for each element against the max values obtained last season - i.e. for PPG I max it out at 30, for APG maxed at 12, SPG maxed at 3, RPG maxed at 15 and BPG maxed at 3. Then assign scores as per element. Found that if we make use of all 5 elements - all players' values get depressed (as invariably there will be at least one area that they lack severely). So decided to go with a model of best 3 scores - but I noticed that for some of the superstars it's still too low - so did another twist to the scores - I magnified their scores in their areas that they weren't expected to perform well (i.e. for forwards I x2 the scores received for APG and SPG and for guards they get BPG and RPG x2). So here's how the guards and forwards rank (well at least I got some of the superstars into the hopefully correct positions. One thing of note though - it seems with the new method Forwards seemed to fair better than guards (but I guess that is alright since Forwards are more than adequately sized to do a guard's job but not necessarily vice versa). So here goes the new baseline ranking (based on Adjusted Best 3 averages or AB3): Guards: 1. R. Rondo Celtics 1972 (note - this was the last known stats and not 2012-2013) 2. K. Bryant Lakers 1878 3. C. Paul Clippers 1863 4. J. Harden Rockets 1845 5. P. George Pacers 1842 6. R. Westbrook Thunder 1818 7. D. Wade Heat 1755 8. R. Rubio Timberwolves 1702 9. S. Curry Warriors 1644 10. M. Ellis Mavs 1635 11. D. Rose Bulls 1628 12. J. Holiday Pelicans 1599 13. G. Vasquez Kings 1574 14. J. Wall Wizards 1572 15. K. Irving Cavs 1539 16. K. Walker Bobcats 1522 17. T. Parker Spurs 1512 18. M. Conley Grizzlies 1499 19. D. Williams Nets 1480 20. K. Lowry Raptors 1443 21. B. Jennings Pistons 1442 22. G. Dragic Suns 1427 23. J. Nelson Magic 1421 24. T. Lawson Nuggets 1420 25. D. Lillard Blazers 1412 26. J. Teague Hawks 1383 27. T. Allen Grizzlies 1378 28. J. Lin Rockets 1302 29. D. DeRozan Raptors 1276 30. M. Carter-Williams 76ers 1271 (P) 30. B. McLemore Kings 1271 (P) 30. T. Burke Jazz 1271 (P) 33. K. Thompson Warriors 1268 34. E. Bledsoe Suns 1230 35. G. Hill Pacers 1223 36. R. Felton Knicks 1220 36. S. Nash Lakers 1220 38. L.J. Mayo Bucks 1211 38. G. Henderson Bobcats 1211 40. B. Beal Wizards 1205 41. T. Sefolosha Thunder 1192 42. D. Green Spurs 1186 43. A. Afflalo Magic 1184 44. W. Matthews Blazers 1175 45. J. Calderon Mavs 1166 46. J. Richardosn 76ers 1153 47. J. Johnson Nets 1122 48. E. Gordon Pelicans 1073 49. J. Butler Bulls 1066 50. D. Waiters Cavs 1047 51. L. Williams Hawks 1025 52. M. Chalmers Heat 999 53. J. J. Redick Clippers 995 54. K. Martin Timberwolves 989 55. L. Ridnour Bucks 970 56. A. Bradley Celtics 957 57. I. Shumpert Knicks 897 58. A. Burks Jazz 689 59. C. Billups Pistons 653 60. E. Fournier Nuggets 489 Forwards 1. L. James Heat 2400 2. K. Durant Thunder 2299 3. D. Howard Rockets 2045 4. A. Iguodala Warriors 1995 5. J. Noah Bulls 1946 6. P. Pierce Nets 1902 7. D. Cousins Kings 1883 8. J. Smith Pistons 1856 9. N. Batum Blazers 1853 10. G. Monroe Pistons 1850 11. T. Duncan Spurs 1848 12. T. Evans Pelicans 1796 13. B. Griffin Clippers 1790 14. L. Sanders Bucks 1777 15. R. Gay Raptors 1768 16. A. Bynum Cavs 1758 16. K. Love Timberwolves 1758 18. A. Horford Hawks 1744 19. C. Anthony Knicks 1714 20. A. Davis Pelicans 1706 20. D. Lee Warriors 1706 22. M. Gasol Grizzlies 1695 23. T. Young 76ers 1689 24. S. Ibaka Thunder 1656 25. A. Jefferson Bobcats 1645 26. L. Aldridge Blazers 1621 27. P. Millsap Hawks 1607 28. L. Deng Bulls 1599 29. C. Parsons Rockets 1585 30. Z. Randolph Grizzlies 1580 30. E. Turner 76ers 1580 32. R. Hibbert Pacers 1572 33. B. Lopez Nets 1563 34. N. Vucevic Magic 1555 35. D. West Pacers 1547 36. K. Garnett Nets 1542 37. K. Leonard Spurs 1528 38. C. Boozer Bulls 1525 39. P. Gasol Lakers 1514 40. A. Drummond Pistons 1509 41. S. Marion Mavs 1492 42. K. Faried Nuggets 1473 43. D. Favors Jazz 1454 44. A. Johnson Raptors 1449 45. C. Bosh Heat 1427 46. N. Pekovic Timberwolves 1419 47. D. Gallinari Nuggets 1413 48. G. Davis Magic 1408 49. M. Gortat Suns 1394 50. N. Hilaro Wizards 1391 51. G. Hayward Jazz 1369 52. O. Asik Rockets 1353 53. E. Ilyasova Bucks 1350 54. D. Nowitzki Mavs 1337 55. T. Thompson Cavs 1326 56. T. Chandler Knicks 1323 56. A. Bogut Warriors 1323 58. J. Dudley Clippers 1282 59. K. Olynyk Celtics 1271 (P) 59. O. Porter Wizards 1271 (P) 59. C. Zeller Bobcats 1271 (P) 62. D. Jordan Clippers 1214 63. K. Korver Hawks 1200 64. E. Okafor Wizards 1192 65. S. Hawes 76ers 1186 66. T. Splitter Spurs 1178 67. J. McGee Nuggets 1140 68. R. Lopez Blazers 1134 69. J. Thompson Kings 1102 70. J. Green Celtics 1088 70. M. Morris Suns 1088 72. T. Prince Grizzlies 1077 73. K. Perkins Thunder 1066 74. M. Kidd-Gilchrist Bobcats 1055 75. T. Harris Magic 1022 76. J. Valanciunas Raptors 1009 77. A. Bargnani Knicks 987 78. S. DaLembert Mavs 951 79. N. Young Lakers 946 80. C. Butler Suns 940 81. E. Clark Cavs 938 82. C. Kaman Lakers 921 83. L. Mbah a Moute Kings 916 84. B. Bass Celtics 905 85. C. Budinger Timberwolves 872 86. E. Kanter Jazz 760 87. J. Smith Pelicans 749 88. U. Haslem Heat 730 89. J. Henson Bucks 722 90. D. Granger Pacers 585 One off-shoot of this adjustment is that C. Parsons ranking value went up. But Asik and Lin's ranking went down. I am likely to go with this baseline for starters - but comments and suggestions are always welcomed.
Become a contributing member and you can edit, among other things. http://www.clutchfans.net/tipjar.cfm
Nice effort, but I do not think that it is helpful to double the unexpected areas. Nor should we just pick the top 3 stats--that would boost certain players that only do well in 3 categories but horribly in the other two. Doing this can also swing the ratings for those swingmen that happen to be "mislabeled" as a G instead of a F or vice versa. What you are trying to do is create a more balanced and better rating system than PER, which I wholeheartedly support.
The new baseline definitely does better with an eye test that the previous one. Bron and KD so low in the original baseline was ridiculous. From the eye test of the top 10 in each category, the ones who seem too high are Rondo, Rubio, Iggy, Noah. Observations of these players that may uncover what needs to be changed. - All 4 are generally weak scorers - Rondo/Rubio - both have abnormally high assists vs the rest of the field - Iggy/Noah - Obnormally high in all stats except PPG (remember Noah's assist and steals for his position). Maybe I'm being hard on these 2, but they are in front of guys like TD. I don't think you need to change the PPG weighting, because then you'll have the K-Marts and Monta's of the world being overvalued, but can offensive efficiency be incorporated? Even something simple like FG% or TS%?
