The probabilites will probably not operate independently in this way. If your hand is good enough to beat the second player, it is likely to be good enough to beat the first player too. So while it depends on the hand, I'm going with something closer to 20%.