Since pawn chess is a competitive game and the players have perfect informatione as well as the game states are simple for representation (discrete), I implemented a Minimax algorithm. While simulating the game for each step to a certain level, the moves are awarded a score depending on how this move influences the outcome.
In my solution I have not considered the probability that the current player of a move can play again. First I had no clue what strategy one should apply for a certain probability value. Second I ran out of time for implementing a solution.
I did not consider measuring time in my solution either, since my algorithm should end before time is up for sure. If evalution is done on a similar or better machine (what I hope so).