OK - then make these guys play 100 or 1000 games against each other and see if anything beyond the very small sample size is responsible for the difference in w

OK - then make these guys play 100 or 1000 games against each other and see if anything beyond the very small sample size is responsible for the difference in win. If the previous winner is superior at RPS he should show a statistical advantage over the loser.