APBRmetrics

Neil Paine · Joined: 13 Oct 2005 Posts: 774 Location: Atlanta, GA

Added WARP file to the list.

The method is to replace a player (on an otherwise average 2007 team) with a replacement -- a player with 95.0 ORtg/16.0 %Poss/107.5 DRtg. These values were obtained by looking at the lower 20% of players (in terms of minutes) in every league from 1996-2007. "Skill curve" usage effects (as seen in my team-projection method) were used for the player's teammates when comparing him to the replacement -- so high-usage players "help" their teammates have a higher ORtg by giving them better shots. All stats were normalized to a 106.3 ORtg environment. WARP is the difference between the otherwise-average team's wins with the player and its wins with the replacement.

"Value" is a representation of a fair salary for the player (based on WARP) in 2006-07. Base salary was $750,000; every additional WARP was worth $1,500,000. Here are the most valuable regular seasons since 1978:

Mountain · Joined: 13 Mar 2007 Posts: 1527

DW have you correlated your WARP derived values to actual salary in any way?

Perhaps focusing on most recent season(s) and non-rookie contracts, particularly longterm ones that presumably are value calls moreso than just market prices at that time would yield interesting results. Wonder how the WARP value/salary ratios look for players with similar WARPS but either Above average Offensive ratings or Above Average Defensive ratings or Both. Is the GM bias toward offensive performance evident using this approach and throughout the data or are least the best defenders getting pretty fairly rewarded these days?

Neil Paine · Joined: 13 Oct 2005 Posts: 774 Location: Atlanta, GA

Mountain · Joined: 13 Mar 2007 Posts: 1527

Thinking a little further about my suggestion, I sorted the 118 on the list that earned a $6+ million WARP based value and found 71 over 110.3 on Offensive Rating and 47 below. On Defensive rating only 37 were 102.3 or below while 81 were higher / less strong, with a league rating at 106.3 making the groups 4 points better than average. Among the players with highest value, strong offensive players are more common but that is to be expected as Offensive Rating tracks the individual and has more likelihood to exceed the average than this team stat based Defensive Rating does. To assess pay for offensive vs defensive qualty evenly, alternative more individual player focused defensive measures DCS or defensive adjusted +/- would be preferrable. (The fair salary work at 82 games for the 05-06 season is an existing database for value/cost trend analysis.)