APBRmetrics

I heard something on the radio recently that struck me. Some scientists said they spent a lot of time on trying to de-bias their analysis. Anyone have thoughts about doing this rigorously with basketball analaysis or have good readable links on the topic?

While thinking about this, I wondered if the computer / video programming is at the ready to fairly easily remove just the personally identified features of basketball players (faces, jersey designs / colors, numbers) from video so that players and plays could be evaluated without it being completely obvious who was involved (except for at least a few players with trademark styles). Would this be feasible in terms of time and cost for large quantities of plays?

Simplier, I can see more merit to reviewing player data temporarily blind to their name.