Great work on the project! You have a very clearly defined and strongly motivated problem and present an interesting solution. The description of the algorithms for each of your approximations was very nice. However, more explanation of some of the subroutines would be helpful in understanding exactly how the implementation works. The experiments themselves were designed well, but it was not clear to me what your baseline "Test" value is--is this the result produced by the builtin AI? Or your human player tests? It is also a bit vague how you chose these army compositions for the experimental games. Finally, it would have been nice to see some theoretical analysis of these results--are any of them admissible heuristics for estimating the next state (i.e., which army will win and number of units remaining)? Overall very nice job with some interesting questions for future work. Grade: A