Multi-objective Optimization of Investment Strategies Based on Evolutionary Computation Techniques, in Volatile Environments

摘要

In this document, the use of a multi-objective evolutionary system to optimize an investment strategy based on the use of Moving Averages is proposed to be used on stock markets, able to yield high returns at minimal risk. Fair and established metrics are used to both evaluate the return and the risk of the optimized strategies. The Pareto Fronts obtained with the training data during the experiments conducted outperform both B&H strategy and the classical approaches that consider solely the absolute return. Additionally, the PF obtained show the inherent trade-off between risk and returns. The experimental results are evaluated using data coming from the principal world markets, namely, the main stock indexes of the most developed economies, such as: NASDAQ, S&P500, FTSE100, DAX30 and NIKKEI225. Although, the experimental results suggest that the positive connection between the gains with training and testing data, usually assumed in the single-objective proposals, is not necessarily true for all cases.