In this paper we estimate the distribution of the initial predictions of the Heemeijer et al. (2009) Learning-to-Forecast experiment. By design, these initial predictions were uninformed and hence one would expect them to follow a continuous distribution. We show that the initial forecasts have a non-continuous distribution and that they systematically under-evaluate the fundamental price.

Our conclusions are based on Diks et al. (1996) test which measures the proximity of two vector sets even if their underlying distributions are non-continuous. We show how this test can be used as a fitness for Genetic Algorithm optimization procedure. The resulting methodology allows for fitting non-continuous distribution into abundant empirical data.