A novel dynamic asset allocation system using Feature Saliency Hidden Markov models for smart beta investing

  • Elizabeth Fons1,2
    •
  • Paula Dawson2
    •
  • Jeffrey Yau3
    •
  • Xiao-jun Zeng1
    •
  • John Keane1
    •

     

     


     

    Proposed Dynamic Asset Allocation system. We design a dynamic trading framework where each day a new vector of returns is added to the training set with an expanding window, and the state is predicted. Because this prediction is noisy, we determine an optimal window of consecutive days (d) in the new state before the portfolio is re-balanced. Once a change of state has been accepted, the vector of means and covariance matrix from the new state are retrieved and the portfolio weights optimized, with transaction costs calculated after the rebalance. After a full month has passed, we add this new batch of data to the training set with an expanding window and retrain the model.

     

    Abstract

    The financial crisis of 2008 generated interest in more transparent, rules-based strategies for portfolio construction, with smart beta strategies emerging as a trend among institutional investors. Whilst they perform well in the long run, these strategies often suffer from severe short-term drawdown (peak-to-trough decline) with fluctuating performance across cycles. To manage short term risk (cyclicality and underperformance), we build a dynamic asset allocation system using Hidden Markov Models (HMMs). We use a variety of portfolio construction techniques to test our smart beta strategies and the resulting portfolios show an improvement in risk-adjusted returns, especially on more return oriented portfolios (up to 50% of return in excess of market adjusted by relative risk annually). In addition, we propose a novel smart beta allocation system based on the Feature Saliency HMM (FSHMM) algorithm that performs feature selection simultaneously with the training of the HMM, to improve regime identification. We evaluate our systematic trading system with real life assets using MSCI indices; further, the results (up to 60% of return in excess of market adjusted by relative risk annually) show model performance improvement with respect to portfolios built using full feature HMMs.

     

    Predicted state probabilities for an HMM trained with three smart beta factors (Book-to-Price, Momentum and Return-on-Equity) and their corresponding cumulated returns. We can observe that the HMM is able to capture meaninful regimes e.g. it starts to transition to a high volatility state at the end of 2007 and is able to detect the 2008 crisis. We can also observe different behaviours on the smart beta factors depending on the regime.

     

    Paper

     

    Bibtex

     

    Acknowledgement

    The authors are grateful to Sahil Kahn, David Hutchins and Andrew Chin for their valuable feedback on early results of this work. This work was supported by the European Union's Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie Grant Agreement no. 675044 (http://bigdatafinance.eu/), Training for Big Data in Financial Research and Risk Management.