Algorithmic Portfolio Rebalancing

This thesis aims at testing the Efficient Market Hypothesis (EMH) by implementing and evaluating four distinct algorithms (Universal Portfolio, Exponentiated Gradient, Anticor and Constant Proportion Portfolio Insurance) for automated rebalancing of fixed-asset portfolios based on the past performance of the individual assets included in the portfolio. If the EMH holds, technical analysis such as algorithm based investments should not be able to generate abnormal returns without introducing abnormal risk. The algorithms are implemented according to the articles presenting them, and I perform statistical hypothesis tests to determine whether the algorithms can provide significant positive abnormal return over broad indices. The results indicate that abnormal returns above broad indices such as the S&P500 and the STOXX are possible. In Monte Carlo simulations, results are statistically significant at the 1% level for three of the algorithms. In tests on actual time series, one algorithm provides statistically significant abnormal return on the 5% level. All algorithms have economically significant abnormal returns for actual time series, indicating that technical analysis can create value despite the assertion of the weak form of the EMH. Trading costs are also introduced. Two algorithms prove very sensitive to trading costs, one is fairly sensitive and one is almost completely insensitive to trading costs, indicating that this algorithm might be useful even for smaller investors…

Author: Mattias Lundahl

Source: Stockholm School of Economics

Leave a Comment