Really like that adjusted scores lean on success rate and play down turnovers and fumble luck. Is there a point in the season where they start to clearly beat actual scores as a predictor, or is the edge fairly steady from Week 4 on?
The edge is actually higher earlier in the season, when the randomness of those high variance plays is greater. As the season progresses, stuff tends to regress, plus there are a lot more qualitative factors to incorporate: injuries, coaching changes, etc
That makes sense: early on there's more noise to strip out, and later the news side matters more. Thanks Kevin!
Really like that adjusted scores lean on success rate and play down turnovers and fumble luck. Is there a point in the season where they start to clearly beat actual scores as a predictor, or is the edge fairly steady from Week 4 on?
The edge is actually higher earlier in the season, when the randomness of those high variance plays is greater. As the season progresses, stuff tends to regress, plus there are a lot more qualitative factors to incorporate: injuries, coaching changes, etc
That makes sense: early on there's more noise to strip out, and later the news side matters more. Thanks Kevin!