Factor tilting for expected utility maximization

We investigate a strategy of investing in diversified portfolios with a historically optimal factor profile, which we refer to as 'factor tilting'. The proposed approach approximates the optimal strategy for risk-averse investors under the assumptions of Arbitrage Pricing Theory. Moving beyond traditional mean-variance optimization, it allows the incorporation of any characteristic of the return distribution for a large number of stocks. We propose extensions to incorporate transaction costs and test factor significance.