Page 1 of 1

Looking For Opinons On Offensive Metric

Posted: Fri Apr 12, 2013 12:58 am
by Jacob Frankel
So I've been working on an offensive metric for the past month of so and thought I'd through it out here for some feedback.

I won't go into the specifics about the methodology, but it's based on the boxscore (with some things like three point assists) and expected points per possession values. It's calculated on a per possession basis and then standardized. 0 is an average offensive player, above 15 is really good, below -15 is really bad.



Note that the data is of around a month ago. Thanks in advance for any feedback.

Re: Looking For Opinons On Offensive Metric

Posted: Fri Apr 12, 2013 8:52 pm
by v-zero
First thoughts:

Not enough PGs at the top, not enough weight to three point wing scorers (often PGs now too).

Re: Looking For Opinons On Offensive Metric

Posted: Fri Apr 12, 2013 10:38 pm
by Jacob Frankel
Figuring out the value of lower usage players who hit threes is probably the toughest part (And getting the right value for assists is ambiguous as well).

On the team level, this metric has an R^2 value of 0.88 when regressed against points per possession and most of the difference I'm chalking up to the different value of assists on the individual and team levels. A little bit more about the methodology: I use the xPPP values to add (subtract) when possessions are gained (lost). I add up marginal points produced through points, assists, offensive rebounds etc. and points lost through turnovers, missed shots etc. Updated version of the chart: