Bucks Diary

Friday, November 09, 2007

NBA Power Rankings by PVOA

I wanted to determine the relative strength of each NBA team thus far, regardless of schedule. So I borrowed the thinking behind the DVOA or "defense adjusted value over average" rankings used by footballoutsiders.com.

DVOA ranks football teams according to how their performance compared with what the average team would have done in a similar situation. Using the same thinking, I ranked the performance of NBA teams thus far through PVOA, or "point value over average", which is meant to calculate how each NBA team is doing so far as compared to how an average NBA team would be expected to perform given a similar set of opponents.

PVOA methodology

To calculate PVOA, what I did was I first calculated each team's Pts per 100 possessions for both offense and defense. Then I charted the score and the number of possessions for each team for each NBA game played through yesterday. Then I compared how the number of points each team scored and gave up in each game compared with the average number of points their opponent would have scored and would have allowed given the same number of possessions, and then I assigned the team a positive or negative amount for points accordingly. Finally, I added together all such scores for that team's games played thus far, and assigned the team a positive or negative average above value for the season.

Here's how it works. If Toronto played Milwaukee, and Toronto scored 100 points in 95 possessions while Milwaukee scored 102 points in 96 possessions, in calculating Toronto's PVOA for that game, I calculated what Milwaukee would have given up and would have scored on average using Milwaukee's offensive and defensive pts per 100 possessions numbers.

For instance, let's say on 95 possessions the Bucks on average would be expected to give up 104 points, and on 96 possessions they would be expected to score 104 points. Then, since Toronto scored 100 points and gave up 102 that game, then for that particular game Toronto's offense would be a -4.0 points over average, and its defense would be a -2.0 points over average, and thus Toronto would get a cumulaitve -2.0 PVOA for the game (for offense, obviously, positive is good while for defense negative is good). In other words, they came up net 2.0 points short of what the average team would have had for its net.

Interesting Results

Since this ranking system is based purely on performance against average, and not at all on my subjective beliefs about which teams are better than others, it yields some interesting, counterintuitive results. For instance, I have the Lakers as the number 2 team in the NBA, even though they are only 2-2, while I have New Jersey as the number 24 team even though they have a much better record. The reason is that the Lakers have either scored or prevented points (or both) at a consistently above average rate but still lost games (probably due to a combination of misfortune and strength of schedule), whereas the New Jersey Nets have done the opposite. And the thinking behind DVOA, and PVOA for that matter, is if you are outperforming or underperforming association averages, as the Lakers and Nets currently are, your win/loss total should reflect that over time.

Other observations about the rankings

The Atlanta Hawks are playing outstanding basketball on both sides of the court, and their fourth place ranking reflects that. Meanwhile, the big 3 in the West are down a bit, as the Suns simply haven't been up to par, neither have the Spurs, and the Mavericks are outstanding on offense, but thus far putrid on defense. But, caveat emptor. Since it is early in the season, these rankings are very fluid. For instance, the Pistons had a much higher average until last night when they allowed the Bulls, a terrible offensive team, to score +12 points above average while they themselves were held -4 below average by the Bulls. The Bulls, meanwhile, have been playing outstanding defense, but have been far below average on offense. Once they get their offense going, as they proved last night, the wins should come.

Here are the NBA PVOA Power Rankings thru Thursday, Nov. 8:

1. Boston......+14.97
2. LA Lakers.......+9.43
3. Houston......+9.39
4. Atlanta......+8.90
5. Detroit......+6.96
6. Toronto......+6.62
7. Dallas......+5.59
8. New Orleans......+5.51
9. Orlando.......+4.45
10. LA Clippers.......+3.45
11. San Antonio......+2.08
12. Denver......+0.92
13. Phoenix......+0.85
14. Washington......+0.10
15. Utah......+0.07
16. Memphis......-0.17
17. Indiana......-0.39
18. Philadelphia......-0.54
19. Portland......-1.83
20. Cleveland......-2.13
21. Milwaukee......-2.75
22. New York......-3.48
23. Chicago......-3.93
24. New Jersey......-5.42
25. Miami......-6.48
26. Minnesota......-6.79
27. Golden State......-6.87
28. Sacramento......-7.16
29. Seattle......-11.72
30. Charlotte......-13.41

Tomorrow, I will rank each team according to the best offenses and defenses using the PVOA system.

3 Comments:

At November 9, 2007 at 2:38 PM, Blogger iwatchthenba said...

Where did you get your averages from? Last season? Group of seasons? Just average all current data together? I'm curious.

 
At November 9, 2007 at 4:09 PM, Blogger Ty Will said...

I took all of the possession numbers each team accumulated this season, because I figured no team is exactly the same as it was last season, so last year's possession numbers wouldn't provide much of a comparison.

Here's the problem, though. Taking the numbers from this year involves that Einstein relative motion problem that no one can explain. The numbers I am comparing each team's play against are affected by that team's play, and vice versa.

Take the recent Bucks-Toronto game for instance. The Bucks destroyed the Raptors, who, until that game, had looked very solid.

Well, in evaluating the Bucks performance in that game, I compared Milwaukee's offensive and defensive output against the numbers Toronto's opponents had amassed as of Thursday. But, of course, by doing that I was penalizing the Bucks because they lowered the Raptors numbers substantially. In other words, Toronto's comparative numbers were much better going into the game than coming out, and I was judging the Bucks performance based on the numbers Toronto had coming out.

But I figure as the season wears along the effect of one particular game on a team's overall numbers will diminish to the point of irrelevance.

 
At March 7, 2017 at 9:48 PM, Blogger Unknown said...

jordans for cheap
cheap basketball shoes
jordan shoes
nike air max thea
lacoste online shop
nike air max
nike air huarache
adidas nmd
nike zoom
cheap jordans

 

Post a Comment

<< Home