Trend-Following, Risk-Parity and the Influence of Correlations
UBS Investment Bank; Queen Mary, University of London; Imperial College Business School
October 12, 2015
“Risk-Based and Factor Investing”, Quantitative Finance Elsevier/ISTE, 2015 (Forthcoming).
Trend-following strategies take long positions in assets with positive past returns and short positions in assets with negative past returns. They are typically constructed using futures contracts across all asset classes, with weights that are inversely proportional to volatility, and have historically exhibited great diversification features especially during dramatic market downturns. However, following an impressive performance in 2008, the trend-following strategy has failed to generate strong returns in the post-crisis period, 2009-2013. This period has been characterised by a large degree of co-movement even across asset classes, with the investable universe being roughly split into the so-called Risk-On and Risk-Off subclasses. We examine whether the inverse-volatility weighting scheme, which effectively ignores pairwise correlations, can turn out to be suboptimal in an environment of increasing correlations. By extending the conventionally long-only risk-parity (equal risk contribution) allocation, we construct a long-short trend-following strategy that makes use of risk-parity principles. Not only do we significantly enhance the performance of the strategy, but we also show that this enhancement is mainly driven by the performance of the more sophisticated weighting scheme in extreme average correlation regimes.
Trend-Following, Risk-Parity And The Influence Of Correlations – Introduction
Trend-following is a simple trading strategy that consists of long positions for upward trending assets and short positions for falling assets. This strategy profits when assets continue performing in-line with their most recent performance. In other words, this strategy aims to take advantage of return autocorrelation empirical patterns.
Trend-following strategies are largely employed by systematic funds, like commodity trading advisor (CTA) and managed futures funds (see Covel, 2009 for a broad overview), and are typically constructed using futures contracts across all asset classes in an effort to increase diversification. The benefit from using futures contracts is two-fold: first, taking long and short positions using futures contracts is equally straightforward (in contrast, for instance, to using cash equity instruments) and second, the use of futures contracts allows the inclusion of non-equity contracts in the portfolio (e.g. trading commodities for investment purposes is typically done using futures).
The construction of a trend-following portfolio involves an important challenge, which is the choice of the weighting scheme that should be employed, given that contracts from different asset classes have very different risk-return profiles (a typical commodity or equity index contract is much more volatile than a government bond contract). An equal-weight allocation would result in a portfolio that would be dominated in terms of risk by the higher volatility assets, i.e. equities and commodities. Instead, the weighting scheme should make use of the relative riskiness of the contracts in order to allocate risk as evenly as possible across all constituents.
The typical choice is to employ inverse-volatility weights, so that all assets enter the portfolio with the same ex-ante volatility. For obvious reasons, this scheme is known as the volatility-parity scheme. This approach has been followed by the large majority of academic research papers focusing on the topic: e.g. Moskowitz, Ooi and Pedersen (2012), Hurst, Ooi and Pedersen (2012, 2013) and Baltas and Kosowski (2013, 2015). Importantly enough, as long as all pairwise correlations are equal, this weighting scheme splits the total portfolio volatility equally across all portfolio constituents.
Using a broad dataset of 35 futures contracts from all asset classes (energy, commodities, fixed income, foreign exchange and equities) we construct a volatility-parity trend-following strategy and document its superior performance relative to a long-only equivalent over a long history of more than 25 years (1988 to 2013). By employing long and short positions, the trend-following strategy benefits from (either upwards or downwards) trending markets and achieves in neutralising (at least unconditionally) the exposure to standard benchmark indices like the MSCI World Index or the S&P GSCI Index. The strategy benefits from the combination of different asset classes and delivers a Sharpe ratio of 1.31 compared to a 0.70 for the long-only equivalent over the entire sample period.
Contrary to the historical superior performance and following an impressive performance in 2008, the trend-following strategy has consistently delivered very poor performance in the post-crisis period (see also Hurst, Ooi and Pedersen 2012 and Baltas and Kosowski 2013). Between January 2009 and December 2013, a volatility-parity trend-following strategy delivers a Sharpe ratio of 0.31 against a Sharpe ratio of 0.59 for the long-only counterparty. What could have possibly gone wrong?
Following the introduction of the Commodity Futures Modernization Act (CFMA) in 2000, commodities have started becoming more correlated to each other as futures markets became accessible to investors as a way to hedge commodity price risk in what is often referred to as the “financialisation of commodities”.6 More generally and more aggressively, following the recent financial crisis in 2008, assets from different asset classes (and not just commodities) have started exhibiting stronger co-movement patterns, with the diversification benefits being dramatically diminished.
In an environment of increased asset co-movement, the volatility-parity weighting scheme can be deemed a suboptimal choice. By ignoring the covariation between assets, volatility-parity fails to allocate equal amount of risk to each portfolio constituent. This is the reason why volatility-parity is also often called as naïve risk-parity (Bhansali, Davis, Rennison, Hsu and Li, 2012). Following these observations, one possible reason for the recent lacklustre performance of trend-following can be the suboptimal weighting scheme that ignores pairwise correlations (see e.g. Baltas and Kosowski, 2015). Our aim is to address this particular feature of the strategy and construct a portfolio that formally accounts for pairwise correlations.
See full PDF below.