On the alternative hypotheses for the win ratio