Comment on “An algorithm for moment-matching scenario generation with application to financial portfolio optimization”

A paper by Ponomareva, Roman, and Date proposed a new algorithm to generate scenarios and their probability weights matching exactly the given mean, the covariance matrix, the average of the marginal skewness, and the average of the marginal kurtosis of each individual component of a random vector. In this short communication, this algorithm is questioned by demonstrating that it could lead to spurious scenarios with negative probabilities. A necessary and sufficient condition for the appropriate choice of algorithm parameters is derived to correct this issue.