Greetings. I am trying to automate some Game Theory strategies (http://www.advancednflstats.com/2008/08/game-theory-and-fantasy-draft-strategy.html) which require the prediction of the positions (QB, WR, TE, DEF, K) your opponents will do between your own. For example, in a 10 team league snake format, you get the #4 overall pick. A snake format via Wikipedia is where "the draft order from the first round is reversed in the second round, then reversed again in every round thereafter, thus creating a snakelike progression" . Your next pick would then be the #17th overall, and of the 13 picks that pass, 9 will be RBs and 4 will be WRs.

Given a number of complete mock drafts for training, I want to be able to predict your opponents likely position picks from any point of the draft. The way I see it, its like weather prediction, where you have N teams inputs for their current state (what positions they drafted already) to find the desired output using historical data. I have no clue how to implement this problem, or if there are any similar problems in computer science that I can model.

Any ideas?