Sample Size and the Accuracy of the Generalized Lambda Distribution

The Generalized Lambda Distribution is a popular tool for generating random numbers following a wide range of non-normal, asymmetric distributions. In this short paper we assess the accuracy of the technique in replicating moments across a range of “sample sizes.” We find that accuracy is highly dependent on sample size, particularly with respect to the third and fourth moments. Generally, acceptable accuracy is accomplished with no less than 2,500 observations, but this finding depends critically on the size of the fourth moment.