Pruning generalized rules for stock markets based on genetic algorithm

This paper proposes a new strategy on pruning generalized multi-order rules accumulated by Genetic Network Programming with Rule Accumulation (GNP-RA). In the pruning method, the usage of each rule (flagged by variable U) and the number of the days having important information in each rule (flagged by variable N) can be evolved by GA. As a result, the pruned rules with better combinations of variable U and variable N are obtained by the crossover and mutation of these variables. The proposed method is verified through experimental studies in stock markets. The effectiveness and efficiency of the proposed method are proved by simulation results.