I’m trying to come up with a formula to predict, as best as possible, how many wins an NBA prospect would likely produce in his rookie season in the NBA based upon the number of wins he produced in his final NCAA season.

So I went back in time and compared about 100 players final collegiate MWS40, and its concomitant winning percentage, against the winning percentage the player posted in his rookie campaign.

The regression equation I came up with was:

##### Y= .17 + .41X

Where Y is the player’s expected number of wins produced in a given amount of playing time, and X is the player’s expected number of wins produced if he posted the same winning percentage as he posted in his final college season.

I then went back and tested how predictive this formula would have been had it been applied to last season’s first round picks (plus DeJuan Blair). The predicted number, based on the formula, had a 89% correlation with the actual number. So the relationship between the predicted number and the actual number of wins produced in the player’s rookie season turned out to be rather strong.

I also ran another regression using only the numbers from the 2009 Draft Class (I did not include any of last summer’s class in my initial regression), and the same method as above, and came up with virtually the same formula:

##### Y= .14 + .42X

So predicting initial pro performance can be done with some success using the player’s final MWS40 as a guide post. I will therefore be using the above formula to predict the likely wins produced by the top prospects in this month’s upcoming NBA draft and I will post my results.

Below I list my calculations of the predicted winning percentage for the first rounders from last year’s draft versus their actual winning percentage, as that winning percentage is posted on my 2009-10 NBA Win Charts Page. (Note: **The list below has no particular order. It just follows my stack of notes. The first number in parenthesis is the player’s estimated Marginal Win Score per 40 –MWS40– from his final college season. Second number in the anticipated winning percentage as a rookie based on upon that final MWS40. Third number is the player’s actual rookie winning percentage.**)

#### Projecting Last Summer’s NBA First Round

##### 1 Stephen Curry…..(+2.83)…..(.468%)…..(.549%)

##### 2. Hasheem Thabeet…..(+5.16)…..(.665%)…..(.472%)

##### 3. Toney Douglass…..(+1.92)…..(.390%)…..(.440%)

##### 4. DeMarre Carroll…..(+4.17)…..(.582%)…..(.425%)

##### 5. Eric Maynor…..(+2.31)…..(.423%)…..(.289%)

##### 6. Taj Gibson…..(+1.54)…..(.357%)…..(.465%)

##### 7. Darren Collison…..(+1.95)…..(.394%)…..(.489%)

##### 8. Jrue Holiday…..(+1.49)…..(.353%)…..(.395%)

##### 9. Jeff Teague…..(+2.07)…..(.402%)……(.389%)

##### 10. James Johnson…..(+3.87)…..(.554%)…..(.047%)

##### 11. Austin Daye…..(+2.49)…..(.439%)…..(.395%)

##### 12. Wayne Ellington…..(+3.15)…..(.459%)…..(.258%)

##### 13. Gerald Henderson…..(+1.94)…..(.390%)…..(576%)

##### 14. Tyler Hansborough…..(+3.93)…..(.562%)…..(.192%)

##### 15. Earl Clark…..(+1.45)…..(.349%)…..(.059%)

##### 16. Terrence Williams…..(+4.02)…..(.570%)…..(.395%)

##### 17. Jordan Hill…..(+0.91)…..(.304%)…..(.290%)

##### 18. Jonny Flynn…..(+0.60)…..(.279%)…..(.051%)

##### 19. Tyreke Evans…..(+2.88)…..(.472%)…..(.586%)

##### 20. DeJuan Blair…..(+7.43)…..(.857%)…..(.884%)

##### 21. James Harden…..(+2.23)…..(.419%)…..(.704%)

##### 22. Ty Lawson…..(+5.12)…..(.661%)…..(.741%)

##### 23. DeMar DeRozan…..(-0.01)…..(.226%)…..(.200%)

**Endnote**: The Marginal Win Score per 40 (MWS40) estimations come from two sources: the player’s WS40 posted on draftexpress, and my calculation of the team’s Oppo WS using the Team Pages featured on statsheet.com and then distributing responsibility according to position.

June 17, 2010 at 12:17 am |

“I then went back and tested how predictive this formula would have been had it been applied to last season’s first round picks (plus DeJuan Blair). The predicted number, based on the formula, had a 89% correlation with the actual number. So the relationship between the predicted number and the actual number of wins produced in the player’s rookie season turned out to be rather strong.”

-How did you get 89%? When I did the Pearson correlation between the 2nd and 3rd numbers in your chart, I got 52%!

June 17, 2010 at 3:07 am |

I ran a comparison of predicted wins over the same number of playing minutes, not winning percentage. If you only used winning percentage, the correlation shouldn’t even be that high, I don’t think.

Try it with wins, tell me what you get.

June 17, 2010 at 12:23 pm |

Jerble,

If you could run those numbers and let me know your results, I would really appreciate it. I really need someone to independently look at this stuff.

Here’s how I did it. I just arbitrarily assumed every guy would play 2410 minutes, because that works out to 10 game responsibilities, if you follow the stuff I do on this site. I did that for ease of comparison.

Then I calculated what the guy would be expected to win if he put up the same numbers as he did in college.

Then I took that number and ran it through the formula:

Y= .17 +.42x, with x being that “collegiate” number I was referring to, and Y being the projection.