Quote: Originally Posted By longtallbrad on 10/23/2009
Quote: Originally posted by monkee on 10/23/2009
I was going to make this long interstitial response to naboimp's response to chesters response to etc
but I wont
naboimp allows that 10 games is insufficient sampling - the original point was that 7 games series are crapshoots if 10 games are insufficient sampling
the question is not randomness or no randomness - its a question of degree - there are programming tricks (some of which used to be in place - when this site had clientele - some of which still is (btw)) that can mitigate and moderate outlier activity
the question is whether there is too much randomness - 10 games is too much
To me that depends on what you're trying to establish. 10 games should generally be more than is needed to identify the better of two teams - unless they are extremely closely matched.
I would hope that 10 games isn't ALWAYS all it takes to determine which teams are the cream of the cream out of a field of 24, or else - as naboimp and all3 suggest - the regular season becomes pretty meaningless and boring.
but we agree that 10 games should be plenty to decide issues of clear 'on paper' superiority
and 'on paper' should be the determinant - on paper IRL means less due to subjective factors than it should in the sim since on paper in the sim is based on objective stat expectations we paid for
ideally if teams are closely or equally matched then you are right, 10 games isnt going to be enough nor should it be
but an underlying problem is that often a team superior in certain stat categories loses because of outlier outcomes in those specific categories - the strong rebounding squad loses the battle of the boards to a team that has concentrated its assets elsewhere - the strong defensive squad neither holds its weak defensive opponent down nor benefits with more efficient offense against that weak defense etc etc
if on the other hand those losses came because of the other team's strengths (not its on paper weaknesses) then we wouldnt have a problem we would have a competitive sim where there was more than one way to skin the rabbit which should be the goal really
that's how you keep things interesting not by simply substituting randomness for a justification for unearned parity but rather by balancing the impact and salary valuation of all stats and conditionals
that way the salary cap acts as it should to make drafters pick their poisons and live with them - that's how you get competitive balance