NBA Season Statistics:
Games played (for most teams): 3
Games remaining (for most teams): 69
Percentage of season remaining: 95.8% (or 94.4%)
My point is that the statistical outcomes occurring now may well be noise when this is all said and done. Sure, you can say after three games that Golden State looks terrible and the data bears out why.
Are you willing to bet that teams like Washington, Cleveland (I know about Sexland and I like their squad overall - no shade here), Sacramento and Charlotte will all be more highly ranked offensively than, say the Bucks, Raptors, the 76ers, Nets, Clippers, Pelicans and Celtics after the season? How about at the break? What about at the quarter pole? By the way, I am not saying that the current rankings won't hold - just that given what we know, they are unlikely to remain the same with the fullness of time.
In short, trying to discern anything from three or five or even ten games seems silly. That said, assuming a small sample size distribution will hold as more outcomes are realized isn't a problem strictly associated with sporting events as we have seen of late.