Threshold-Based Portfolio: The Role of the Threshold and Its Applications

Abstract

This paper aims at developing a new method by which to build a data-driven portfolio featuring a target risk-return. We first present a comparative study of recurrent neural network models (RNNs), including a simple RNN, long short-term memory (LSTM), and gated recurrent unit (GRU) for selecting the best predictor to use in portfolio construction. The models are applied to the investment universe consisted of ten stocks in the S&P500. The experimental results shows that LSTM outperforms the others in terms of hit ratio of one-month-ahead forecasts. We then build predictive threshold-based portfolios (TBPs) that are subsets of the universe satisfying given threshold criteria for the predicted returns. The TBPs are rebalanced monthly to restore equal weights to each security within the TBPs. We find that the risk and return profile of the realized TBP represents a monotonically increasing frontier on the risk-return plane, where the equally weighted portfolio (EWP) of all ten stocks plays a role in their lower bound. This shows the availability of TBPs in targeting specific risk-return levels, and an EWP based on all the assets plays a role in the reference portfolio of TBPs. In the process, thresholds play dominant roles in characterizing risk, return, and the prediction accuracy of the subset. The TBP is more data-driven in designing portfolio target risk and return than existing ones, in the sense that it requires no prior knowledge of finance such as financial assumptions, financial mathematics, or expert insights. In a practical application, we present the TBP management procedure for a time horizon extending over multiple time periods; we also discuss their application to mean-variance portfolios to reduce estimation risk.