Conditional Risk-Based Portfolio

Risk-based investment strategies such as Minimum Variance, Maximum Diversification, Equal Risk Contribution, Risk Parity, etc. share the common feature of being based on a risk measure, typically the covariance matrix of the asset returns. When one comes to implement these strategies, the usual approach consists in using an unconditional covariance matrix, simply estimated by the sample covariance matrix of past returns over a rolling window. An alternative consists in using a conditional covariance matrix computed from a multivariate GARCH-type model and which depends on information available to date. In this paper, we propose the first unifying and systematic comparison framework for the unconditional and conditional risk-based investment strategies. We compare their out-of-sample performances in terms of risk, returns and turnover (trading volume) with 4 criteria across 3 empirical datasets. Our results show that conditional risk-based strategies do not improve the out-of-sample Sharpe ratios as well as the ex-post risk, but logically increase the turnover.