Need Help with a Statistic: TPSR

[Nuck Chorris]

So this idea has been itching away at me for a long time. It always pissed me off to see teams say that they are doing their best to be competitive when the reality of the issue is that the money on their books doesn't match the product on the court.

So I decided to create an arbitrary rating that can be used to rank the bottom line effectiveness or productivity of a team's salary obligations in direct relation to the salary cap.

*This stat is far from perfect, and any and all positive contributions will be considered. Your input is greatly appreciated.

Team Productive Salary Rating and Percentage

((Player PER)/(League AVG PER))(Player Salary)= (Adjusted Player Salary)

********* adjusts for individual play. The effect of outliers like Kobe and James is nearly nullified by the low salaries or production of teammates. Keeping in mind that this is supposed to be a reflection of how much the teams are really putting their money where their mouths are (as opposed to the true merit of a single player).

(APS1)+(APS2)+......= (Adjusted Team Salary)

(ATS)((Team w/l %)/(.500)) = TPSR

************ This adjusts for the success of the team. Call it the Battier/Adelman Adjustment.


(TPSR) / (Salary Cap) = TPSP

Or we could do

(TPSR) / (Team Salary) = % of team commitment fulfilled.

Now there are a few conditional variables that come into play here.

- Salary commitments that are paid for by insurance, injury or avoided by suspension are not factored into TPSR or TPSP as they do not actually influence the product on the court.

- Inclusion of veteran minimum contracts (paid by the league) are conditional to a positive sum based on (Player PER) - (Avg. PER). If their PER is negative in relation to the league average then their salary counts against the TPSR.

I wanted to use some version of +/- rather than PER but I am not sure how it would be implemented.

- Just to note with regard to setting a "standard":

A team that spends exactly to the salary cap, has "average" players, and a record of .500 will have a TPSR of 60.7 (exactly the cap) and a TPSP of exactly 100%. 100% meaning that between the play of individual players (reflected in their salary adjustments) and the success for investments/leadership of the team (win %) that the individual team has fulfilled its production commitment (to the mid-point).

I won't give an exact number yet but by my approximation the Warriors' TPSR is in the mid 30's.


Thanks for your contributions.

[Nuck Chorris]

Got my espn scrip. Data will be out in a couple days.

[Alfred]

This seems very interesting. Please keep us in the loop.

[Nuck Chorris]

Player salary will no longer be adjusted or reflected in the statistic as it ultimately has no bearing on the on-court product.

Rather PER will translate directly into PER Adjusted Player Salary (PAPS) by factoring in the MLE relative to the league average PER.

Ultimately I decided to do this because the fans don't OK the money that is signed off to players. So their perception of the on-court product (as represented through these statistics) should not be watered down by inflating or deflating player salaries. Especially when many players are already over or underpaid. Simply too convoluted.

Thanks Cohan. I was tearing my hair out trying to figure out MLE implementation but I just had to refocus the aim of this statistic to pure court production and in that context player salaries become superfluous. However, now the PER Adjusted Player Salary can be compared directly to their actual salary for a by-the-numbers (not actual) assessment of a team's investment. Thank you Hollinger for finally factoring minutes into your PER stat.

Any factoring of minutes into this stat will have to wait until version 2.0, but I really like the idea.

Still crunching numbers. Could take longer than I thought. I'm going to do every team and player.

[Chicago76]

Dan Rosenbaum did something similar using +/- here:

http://www.82games.com/0506/fairsalary.htm

A few suggestions:

1-As you mentioned, you will need to minute-weight the data, otherwise a guy playing a handful of minutes with a PER of 30 will skew your data.

2-If I were you, I would work in some concept of a "Replacement Value" into your salary/value equation. For example, you can pick up a player with a PER of 8-10 for basically the league minimum. The incremental value of a player with a PER of 15 (assuming league minimum salary replacement having a PER of 8) is:

(15-8)*(% of total team floor minutes played) = "PER pts"

Each "PER pt" would be worth something more than the minimum salary. Assume the average team salary is $100, each team has 15 players, and the league minimum salary per player is $1 for simplicity sake. The average team would have a PER of 15. Team "PER pts" = (15-8) x 100% floor minutes or 7 pts above replacement. Those 7 pts would need to be allocated for the salary over league min ($100-$15) or $85. Each PER pt = $12.14.

A superstar with a PER of 30 might be on the court for 90% of a game (or 90% x 1/5 to get team floor minutes). His PER pts would be (30-8)x(.9x.2) = 3.96

His assumed salary = 3.96 PER pts x $12.14/PER pt + $1 =$49.1 The $1 covers the baseline value of a replacement scrub.

[mysticbb]

Still crunching numbers. Could take longer than I thought. I'm going to do every team and player.

I would suggest using b-r.com for that: http://www.basketball-reference.com/pla ... i?id=htOf8

I tried something similar but way more simple. I used Win Shares (because they correlate actually to winning more than PER ;)). A artificial average player would produce 4.1 Win Shares per 82 games (in 24 minutes on a 41 win team). Now I compared it just to the produced Win Shares for each player. Now I say a average player should earn the MLE (5.854m for the this season). That would be the individual "fair" salary for each player. The salary for each team would be much easier. To cover all minutes you would need 10 of those artifical average players. So, a team should spent 58.54m for a 41 win team. Now you compare this to the team salaries (just devide the "fair" salary by their actual salary).

For this season so far (taking the numbers until 01/06/10):

Code: Select all

Oklahoma City Thunder  1.31
Portland Trail Blazers 1.24
Atlanta Hawks          1.16
Cleveland Cavaliers    1.07
Boston Celtics         1.05
Memphis Grizzlies      1.03
Los Angeles Lakers     1.03
Phoenix Suns           0.99
Orlando Magic          0.98
Denver Nuggets         0.98
San Antonio Spurs      0.93
Los Angeles Clippers   0.92
Dallas Mavericks       0.91
Sacramento Kings       0.90
Houston Rockets        0.89
Toronto Raptors        0.86
New Orleans Hornets    0.82
Miami Heat             0.82
Utah Jazz              0.77
Charlotte Bobcats      0.77
Milwaukee Bucks        0.75
Chicago Bulls          0.72
Detroit Pistons        0.65
New York Knicks        0.58
Indiana Pacers         0.57
Philadelphia 76ers     0.54
Golden State Warriors  0.54
Washington Wizards     0.50
Minnesota Timberwolves 0.37
New Jersey Nets        0.17


Now the problem is that the teams are spending an average of 70.29m in this season. Assuming that the income is related to this (which is in fact not the case, last season the players got nothing back from the ESCROW tax!), we can adjust that. That would in fact change nothing in the ranking, but you get a better impression which teams are spending their money like they should (around 1 and above).

Code: Select all

Oklahoma City Thunder  1.58
Portland Trail Blazers 1.49
Atlanta Hawks          1.40
Cleveland Cavaliers    1.29
Boston Celtics         1.26
Memphis Grizzlies      1.24
Los Angeles Lakers     1.23
Phoenix Suns           1.19
Orlando Magic          1.17
Denver Nuggets         1.17
San Antonio Spurs      1.12
Los Angeles Clippers   1.11
Dallas Mavericks       1.10
Sacramento Kings       1.08
Houston Rockets        1.07
Toronto Raptors        1.03
New Orleans Hornets    0.99
Miami Heat             0.98
Utah Jazz              0.93
Charlotte Bobcats      0.93
Milwaukee Bucks        0.90
Chicago Bulls          0.87
Detroit Pistons        0.78
New York Knicks        0.70
Indiana Pacers         0.68
Philadelphia 76ers     0.65
Golden State Warriors  0.65
Washington Wizards     0.60
Minnesota Timberwolves 0.44
New Jersey Nets        0.21

So, right now I would say each team below the Miami Heat are spending too much money for the success they have so far. By a couple of those teams on the top of the ranking you can see the influence of the rookie scale contracts. Some of those players are heavily underpaid. Take Durant, Green and Westbrook as an example. Going by my described method they should earn 33.8m for this season, but in reality they only get 12.1m.