At this point, if it wasn't obvious before, it is clear the NBA wants the Heat in the finals. So you're Stern and you want the Heat in the finals. Well, the Heat can get there in several ways (Game 4, 5, etc.).
Ideally Stern wants them to get to the finals after winning a game 7 vs the Celtics, but unless there is outrageous biased officiating, pushing for a game 7 from your refs would risk the Heat actually losing that game. Nevertheless, if Stern could decide everything to the letter he would have every home game win the current series.
But with the randomness of player performance and ref ability to affect outcomes without being too blatant, how would Stern go about it? I feel like the following is a natural hypothesis:
push refs to ensure Heat win game 1, favor them strongly in game 2 so they probably start off 2/2 and then let the series play out fairly, or maybe officiate in favor of Boston a little. Heat probably take 1 in Boston then favored to close out in Miami. If they don't then have refs work hard to ensure Heat win Game6 in Boston. If Heat go 0/2 in Boston then officials work hard from then on out to ensure Heat win next two games.
This is what we have to begin the series. Now if you have a model that predicts the game 3 outcome, without considering officiating, and has the Celtics favored by x, how would you go about adjusting it to account for NBA tampering?
At this point if the posted line is exactly x, you would have to imagine the Celtics are a strong play since game 3 officiating should be in their favor. The league probably doesn't want a 3-0 lead by the Heat as it would kill interest in the series. But what if the line is x-2? How do you quantify officiating bias, keeping in mind it is not even clear in which direction the bias is pointing. It's only my guess it will favor the Celtics in game 3.
Thoughts??
mathy