The past researches emphasize merely the avoidance of over-learning at the system level and ignore the problem of over-learning at the model level, which lead to the poor performance of the evolutionary computation based stock trading decision-making syst
