Standardizing Jason's Box Score MVG

[themistocles]

Now, although I think that Jason's Box Score MVG is a considerably better way to look at Box Score (and only Box Score) performance, it is biased in favor of those who play more minutes, and at least to this point, he has only computed single game values for players.

So, adopting Jason's method - +5 for Assists & Steals; +4 for Points, Blocks & Rebounds; -3 for Turnovers, -2 for FG attempts; and -1 for FT attempts, I summarized the entire season, came up with a per-minute average for each of these statistics, among others, and computed his scores.

Using a per-minute standardized value, one can compare those who play perhaps 10-15 minutes per game (e.g. Frazier & Prather) with those who play 30 or more (e.g. Boynton). Of course, assuming that what has occurred during a short game stint would continue at that rate for a longer period of time is almost surely, a fallacious assumption, however, everything involves assumptions, and one would hope that idiosyncrasies that occur in specific games would level out over a season.

Now, basing this on the average 20 minutes for each player, I came up with the following season-long values:

60 - Prather
47 - Young
45 - Murphy
40 - Yeguete & Wilbekin
36 - Boynton & Frazier
32 - Rosario
25 - Walker
17 - Ogbueze
14 - Graham
4 - Kurtz

So, Prather, who so far in his 4 games has averaged 12.8 minutes, is performing at a considerably higher rate than any of the other players on the team.

I should note that I cross-verified my numbers against Jason's game by game (he only noted the higher values).

What I found was that in most of the games, our scores for the top players different by from 2 to 5 points. I double checked my values on a couple of games, because perhaps we were using different sources that reported slightly different statistics, but my were identical for both of the Box Score Sources I checked: CBS Sportsline and GameHQ, so, I must assume either that Jason's algorithms are somewhat inconsistent, or, that there are some data entry inconsistencies between the two of us. Mine are, for most things, cut & paste, but, unfortunately, I must hand enter the FG/FT data (UGHHHHH - I hate to hand enter anything - time consuming and a source of error).

They are, however, always, very close.

[gatorbogey]

it's good to normalize the data to help compare value on a per minute or per half basis.

[corpgator]

It's easy to fall prey to the trap of overrating bench players since they spend more time against the other team's bench players.

Usage and efficiency tend to be inverses. The more possessions you use, the less efficient you get. There's also value to a guy who make isolated shots with time running down on the shot clock. His % is probably lower, but it's also higher relative to his teammates taking the same shot.

[themistocles]

Originally Posted by corpgator

It's easy to fall prey to the trap of overrating bench players since they spend more time against the other team's bench players.

Usage and efficiency tend to be inverses. The more possessions you use, the less efficient you get. There's also value to a guy who make isolated shots with time running down on the shot clock. His % is probably lower, but it's also higher relative to his teammates taking the same shot.

These are both excellent points, and yes, both Jason and I know that Box Scores are only one part of the story.

However, I don't think that anyone can say that either Prather or Frazier have played primarily against the bench, and those are the two highest rated Bench players.