I've looked a LOT at building an NBA model, read lots of threads on another forum that is particularly savvy in terms of modelling and American sports. Basically if your looking at line betting whether its the score total or the winning team line then its extremely difficult because so many people are already doing it. So to find a winning edge is next to impossible because just as you think you've found something on one or two years data, it could simply collapse on the next and then its back to the drawing board.

Some of the modelling and statistics knowledge of the guys that do it successfully in the US is simply amazing. Some have gone on to actually work for the NBA teams. I believe one of the guys that posted on the forum and then started his own website eventually couldn't post anymore because he started working for the Denver Nuggets.

Anyhow the best piece of advice that was on the forums was rather than to beat the lines that everyone is betting on (winning handicap and totals) try to find a niche that you can be profitable with on those sports. I.e. single player scoring totals (i.e. Kobe Bryant over/under 23.5 points) or something similar. Typically these lines are underbet and therefore not as 'sharp' as the overbet lines. Also if you shop around on these niche type bets you can very often find a difference between betting agencies for those lines, sometimes 2 or more points, which may help you have just the edge you need to be profitable.

Note the above was my experience from 12 months ago, i've learnt a lot from horse racing since then but haven't attempted to go back to it yet. Basketball and in particular NBA is something I can't get enough of and love this time of year when I sign up for League Pass and watch every game I can.