Genetic Mean Reversion Strategy for Online Portfolio Selection with Transaction Costs

Online portfolio selection (OLPS) is a procedure for allocating assets using only past information to maximize an expected return. There have been successful mean reversion strategies that achieved large excess returns on the traditional OLPS benchmark datasets. We propose genetic strategy evolves population of vectors hybrid algorithm. Each vector represents proportion assets, and our chooses best in terms every trading day. To test strategy, we used price S&P 500 constituents from 2000 2017 compared various online selection. Our framework successfully evolved vectors; therefore, outperformed other when explicit or implicit transaction costs were incurred.