Poll: RPM's degree of efficacy in sorting players

[rlee]

Latest on the issue of how Vegas lines are set:. This is from Jeff Sherman who sets the Westgate's over/under lines:

SN: What kind of factors go into setting the over/under totals?

Sherman: The whole goal is obviously to try and create equal action on the betting number that we put, just like anything else that we take.

So it’s not a predictor of how many wins they’re supposed to get. It’s based on what we think people’s perception is to get two-way volume on that number."

[sndesai1]

if a market is large enough, those two concepts ("prediction", balancing sides) should lead to a very similar point, at least at the books that don't restrict bettors

nba win totals have pretty small limits everywhere though, so it often doesn't

[rlee]

And this explains it in more detail, as it applies to odds & point spreads:

http://www.espn.com/chalk/story/_/id/23 ... ndle-chalk

[Crow]

I can accept the finding that RPM estimates may tend to decline for traded players on average. I would like to see further research at individual (vs. who replaced) & team factors. I heard no indication of research down to that level. I'd also to see how coaching quality affects the transition and whether the loss reduces during season or in next two seasons. By age would be important too. A new team is a new context and can sometimes mean modification of role. Sorting results by degree of similarity of team by efficiencies and / or factors and degree of similarity / change in role could also further understanding. Assuming a decline might be a valid starting point but shouldn't be the end point.

Also, if traded player RPM is less stable than the overall average, it would imply that non-traded player RPM is more stable yr to yr. I wonder if the same is true when role does not change or age / experience is near the middle of the aging curve. Perhaps RPM is a better predictor than it appears overall if the context is better understood / adjusted.

[Parshan]

So, what about a new poll based on different stats?FM WhatsApp

[Mike G]

Go for it!