Read this TrueHoop blog entry, which is an interview with a professional NBA gambler. The Tim Donaghy scandal has gotten pretty interesting over the last couple of days, with public allegations of game-fixing (including the playoffs) coming from the "rogue referee" camp. The timing of this, during the NBA Finals, is, naturally, very embarrassing to the NBA. It’s a very difficult situation for them to handle — if the series goes long, say, the full seven games, it will provide another week and a half of damaging media coverage, plus skepticism about the length of the series being the fault of the officiating. Their best hope, in terms of squashing the story quickly, may be to drop the restitution demand on Donaghy (in return for some sort of retraction), and to hope for a short Finals.
The article also brings up the idea of sabermetrics being used as a tool to "beat the odds" in sports betting. This is not such a far-fetched idea — the oddsmakers aren’t really trying to predict the winners of each game, but rather the handicap that will produce as even a betting spread as possible, to maximize the bookie’s take. Sabermetrics, as an aid to sports betting, would be used to find an edge where the odds set are influenced by "inefficient" or human factors — strong public favor for one team, misleading statistics or hot streaks, etc. — and I could believe that there is likely some exploitable edge there.
I find it interesting that Voulgaris’s NBA betting focuses mainly on the points over-under, and that his analysis (at least according to the article) seems heavily weighted towards defensive matchups and offensive efficiency (in terms of possessions). My guess, not having looked into basketball statistical analysis, is that the overall effect of team defense on total score is more predictable than the effects of star players (who may drastically outperform their averages) on the outcome of a game.