Portfolio Construction and Systematic Trading with Factor Entropy Pooling
Laval University - Département de Finance et Assurance; Centre interuniversitaire sur le risque, les politiques économiques et l'emploi (CIRPÉE)
May 8, 2014
The Entropy Pooling approach is a versatile theoretical framework to process market views and generalized stress-tests into an optimal "posterior" market distribution, which is then used for risk management and portfolio management. Entropy Pooling can be implemented non-parametrically or parametrically. The non-parametric implementation with historical scenarios is more suitable for risk management applications.
Here introduce the parametric implementation of Entropy Pooling under a factor structure, which we name Factor Entropy Pooling. The factor structure reduces the dimension of the problem and linearizes the parameter space, allowing for fast computation of the posterior market distribution.
We apply Factor Entropy Pooling to two portfolio construction problems.
First, we use the Factor Entropy Pooling to construct the "implied returns", i.e. a market distribution consistent with a target optimal portfolio, such as maximum diversification/risk parity, or the CAPM equilibrium. Our approach improves on the implied returns a-la-Black-Litterman, and the ensuing distribution can be used as the starting point for further portfolio construction.
Second, we use Factor Entropy Pooling to construct and backtest quantitative systematic trading strategies based on ranking views, or "portfolios from sorts". Unlike standard approaches, Factor Entropy Pooling closely ties to the actual empirical data.
Fully documented code is available.
Number of Pages in PDF File: 36
Keywords: Trading signals, tactical allocation, Black-Litterman, equilibrium prior, shrinkage, risk management, Entropy Pooling, factor models, inequality views, portfolios from sorts, ranking, Kullback-Leibler
JEL Classification: C1, G11working papers series
Date posted: May 13, 2011 ; Last revised: May 9, 2014
© 2014 Social Science Electronic Publishing, Inc. All Rights Reserved.
This page was processed by apollo7 in 0.265 seconds