Individual Game Rating Metrics

[miabbett]

I'm aware of Hollinger's Game Score, and I recently discovered that ESPN has a variation of metric which they call ESPN Player Rating. Which of these statistics do you think better measure how "good" of a game someone had? Are there any other simple box score based metrics that do a better job of identifying a player's performance in an individual game?

[Mike G]

"Good" might mean either productive - or - efficient.
"Efficient" generally means per-minute or per-possession.
"Productive" is generally a (weighted) sum of totals.

You could take a player's raw +/- for the game, when it's in the box score, and add that to his productions.
You could do this by per-minute or possession.

Invent something!

[Crow]

There are the old NBA efficiency and Berri's winscore http://wagesofwins.com/2007/02/17/using-win-score/ The first may be available at NBA.com or I believe it is as hoopsstats.com. The later isn't at any of the original sources cited in article (all long defunct) but probably is at boxscoregeeks.com or an associated site.

[Crow]

Using a blend of Game Score and Winscore may be better than either alone, as they tend to offset each's distinctive exaggerated credit for usage or rebounding.

Or one could use alternate win score (with a rebounding adjustment. search for it here or on google).

All will lack individual shot defense and therefore be incomplete and often skewed compared to actual total impact. And yet most of those who reject the inaccuracy of rpm will continue to use incomplete / inaccurate roll up boxscore metrics (simple raw or spm) with little or no visible / vocal acknowledgement that they completely lack 20-25% of the game's most consequential action (shot defense).

[miabbett]

Thanks for the responses. Yeah I know it's impossible to truly gauge a player's performance based on box score stats, as well as the fact that it all depends on one's definition of "good." I'm just looking for something that does a decent job recognizing if a guy fills up the stat sheet as well as guys with low minute, high efficiency games. I ended up tweaking the weights of Game Score and changing a few of the variables, which looks halfway decent.

[Mike G]

If you can find a boxscore with Game Score, ORtg, and DRtg, you can try a variant of
GmSc * (ORtg/DRtg)^x

Raise x if you want Efficiency to be more important, relative to Productivity.