Twice before, in 2009 and 2011, I sought to predict the Academy Award winners in six major categories based on a mix of statistical factors. My track record? Nine correct picks in 12 tries, for a 75 percent success rate. Not bad, but also not good enough to suggest that there is any magic formula for this.
So this year, I have sought to simplify the method, making the link to the FiveThirtyEight election forecasts more explicit. …
So our method will now look solely at the other awards that were given out in the run-up to the Oscars: the closest equivalent to pre-election polls. These have always been the best predictors of Oscar success. [cont.]
Nate Silver, FiveThirtyEight (NYT)