Testing the Dinosaur Hypothesis under different GP algorithms

Abstract

The Dinosaur Hypothesis states that the behaviour of a market never settles down and that the population of predictors continually co-evolves with this market. This observation had been made and tested under artificial datasets. Recently, we formalized this hypothesis and also tested it under 10 empirical datasets. The tests were based on a GP system. However, it could be argued that results are dependent on the GP algorithm. In this paper, we test the Dinosaur Hypothesis under two different GP algorithms, in order to prove that the previous results are rigorous and are not sensitive to the choice of GP. We thus test again the hypothesis under the same 10 empirical datasets. Results are consistent among all three algorithms and thus suggest that market behavior can actually repeat itself, and have a number of 'typical states', where past rules may become useful again.