There is a pretty interesting article over at blazer's edge about win margin and win %. http://www.blazersedge.com/2009/8/13/987664/what-did-hollinger-say
I don't see anything in hollingers statement to suggest this at all. Scoring margin is more reliable predictor of future success, not a less biased predictor. For example in an election, a large poll is more reliable than a small poll. But a small poll is not more biased. Both a large poll and a small poll are centered around the true mean. The article is claiming that a team on the rise will have a win% that is biased too low while their point diff will be unbiased or less biased. I believe, not true. What you can expect for a team on the rise is that both point diff and win% will be rising over time, even within a season. Both win% and pt diff will be biased too low as a future predictor because the team is improving.
That is certainly what the little graph of last season split into Total/Last40/Last20 told the reader. Maybe he's just expressing his argument improperly, but trying to make a point like yours above.
Could be, although he makes the same point about win% lagging behind point diff multiple times so I think he's just misinterpreting. I could make one argument in favor of what he says. A young team needs a higher pt. diff to acheive the same win% because an experienced team knows how to win close games. Either way it bodes well for the Blazers, esp. if their end of season pt diff was "real" and given youth improvement plus Sergio --> Miller should provide another boost to that. I'm expecting 60+ wins.
He comes to some incorrect conlusions, but it was an interesting article. You also cannot use last season's scoring margin to predict next season's win total, which I believe he did at the end.
What it seemed to me he was doing was saying, "If Portland has this point differential next season, they should win this many games." You can never literally extrapolate anything directly from past results because everything is changing...but if you were going to make predictions for next season using one stat from last year, point differential would probably be the best one to use. Truly sophisticated models would include some amount of regression to the mean and a way to incorporate the change of personnel and its likely effect on point differential. In any case, it's a reasonable point to make that Portland finished the season with a remarkable point differential in the second half and that if that reflected a "true talent level" (as opposed to variation), then Portland has a chance to be extremely good next season (better than most would expect just from wins). Personnel changes only make the picture rosier, as they added a much better point guard, lost no one of import and their young players will be a year more experienced and closer to their primes.
