The Magic used hack-a-Dwight ("HAD") last night, and it backfired mightily for them. Dwight started 2-10 from the line, and then the Magic started using HAD, and Dwight went on to make 23 of his next 29 free throws.
I think this is an example of one of the problems of the HAD strategy. See, I'll accept that, in games, the distribution of Dwight's free throw percentages will fall around his average. However, something that people haven't considered is that Dwight shoots 80% on free throws in practice. The lakers actually track it and put the tally on a big white board, so this isn't just some story that you hear - it's a fact. To me, this means that in one set of circumstances (game), he shoots 50%. But in another set of circumstance (practice), he shoots a 80%, which obviously is a completely different percentage. When the Magic employed HAD, Dwight made 23-29 free throws. That's 79%, which basically mirrors his practice percentage.
It seems reasonable to assume that the cause of this increased FT accuracy is that the Magic unwittingly created a situation in which Dwight is a good free throw shooter. Essentially, they said, "OK, coin flips land heads 50% of the time. Dwight, take this coin and I'm betting if you flip it 40 times, you'll get heads like 50% of the time." And then they proceeded to give him a weighted coin.
Additionally, if you employ HAD consistently, you give Howard a chance to catch his breath and regain his energy. For a guy who has really struggled with his conditioning this season, giving him an energy booster ever 45 seconds isn't the best way to capitalize on that. It gives him the energy to be a beast on defense. And allowing Kobe and Nash to never have to run past half court on offense is a more effective way of ensuring that they have maximum energy in the tank come crunch time than Mike D'Antoni has come up with.
The main point here is that statistical analysis is going to lead you astray if you only consider consider the most obvious variables and ignore less available, but nevertheless causative, variables.