You can also analyse the skewness and kurtosis from the interval PnL by getting third and 4th times of $Y_t$ respectively. Presumably you might conclude that for 2 collection with similar expectation and variance, you might want the a person with positive skew or lower kurtosis, but maybe not accord