engleski

Portfolio optimization using preference relation based on statistical arbitrage

Abstract: We propose a new algorithm for portfolio optimization based on statistical arbitrage, that uses a multi-criteria decision making approach to obtain the most preferred assets. A preference flow graph of financial assets is constructed at each time step, with the aid of statistical arbitrage algorithm that describes preferences among the assets. Then, the individual preferences for each asset are obtained by using the potential method, and the most preferred assets are selected into the portfolio in accordance to them. A consistency measure of the preference flow graph is also obtained using the same method, and it measures the reliability of the decision making. The proposed method has been tested on a selection of S&P 500 constituent stocks from 1980 to 2004. The results indicate that the proposed method performs well in the considered market, which is indicated by high Sharpe ratios of the constructed portfolios. We also report that the algorithm performs better when provided with a larger number of assets, showing that the increased number of considered assets provides more insight into the market behavior.

Portfolios ; Decision making ; Optimization ; Prediction algorithms

nije evidentirano