Hey Jerry,
There has been some recent twitter talk that RPM is better interpreted as mainly or just fit in context than primarily true skill. I wondered if you had any general reply to that at this time and I have a question:
Could you rate the favorability of a player's average "context" by looking at what the four factors data say about that average context (lineup data or 4 players 4factors summed Wo that player) and post alongside the rpm rating so one could more clearly say this player estimate is plus 2, really that is the best estimate, but it so happens to come within an avg plus 1 or plus 4 or minus 2 context?
Could / would you list all players in league by rank of the temperature of their avg context?
Could you compare how the model estimates players with similar player estimates and contexts, similar and different sets of these and dissimilar on both and discover anything to improve those estimates or the model itself?
And if additional information is shared about average context, would you consider reporting the average rpm estimate of the players substitutes or main substitute as well? If rpm is a resource set of data and not simply a point estimate, maybe that will be more helpful and encourage more "responsible" use?
On context in RPM
Re: On context in RPM
Or really, it might help to say the context is avg. x for teammates and y for avg quality of opponents faced. Will vary for starters / subs, good / bad teams, east / west.