This model generates optimal portfolio weights when the user values more than just mean and variance, but also skewness and kurtosis. It shows the critical importance of co-skewness with the portfolio in achieving ex-post out of sample positive skewness.