An Analysis of the Risk‐Return Characteristics of Serially Correlated Managed Futures

• Date: 01 October 2016
• Published date: 01 October 2016
• Author: Gert Elaut,Péter Erdős,John Sjödin
• DOI: http://doi.org/10.1002/fut.21773

An Analysis of the Risk-Return

Characteristics of Serially Correlated

Managed Futures

Gert Elaut*, Peter Erd}os and John Sj€odin

We investigate the implications of low but persistent serial correlation in Managed Futures’

returns for portfolio management. Using a measure based on the unweighted sum of

autocorrelations, we find that more positively autocorrelated Managed Futures exhibit

distinctly different risk-return profiles and outperform, on a risk-adjusted basis, Managed

Futures that exhibit lower degrees of serial correlation. The observed premium is unlikely to be

explained by a concentration in certain strategies, fund size and age, attrition or delisting

bias, and does not seem to hamper Managed Futures’portfolio benefits as a tail-risk hedge.

© 2016 Wiley Periodicals, Inc. Jrl Fut Mark 36:992–1013, 2016

1. INTRODUCTION

The historical track record remains the most important piece of information in the evaluation

of potential hedge fund managers. This is the case as information on the a-models used by the

managers can only be inferred from their track record. The models themselves remain strictly

proprietary. As a consequence, past returns will remain a key element in manager selection.

An important consideration in this regard, is the degree of persistence in managers’reported

returns. If fund managers’returns exhibit persistence at certain frequencies, then manager

selection based on past performance can potentially add value along this time series

dimension.

In this article, we provide empirical evidence that value can potentially be added

through incorporating serial correlation patterns in Managed Futures’self-reported returns

in the investment process. In particular, we find that Managed Futures funds that

exhibit higher degrees of positive serial correlation—based on the unweighted sum of

Gert Elaut and John Sj€odin are at Department of Financial Economics, Ghent University, Ghent, Oost-

Vlaanderen, Belgium and Peter Erd}os is Investment Analyst at RPM Risk & Portfolio Management AB,

Stockholm, Sweden. Funding from the European Commission’s (EC) Grant Agreement no. 324440

(Futures) Marie Curie Action Industry-Academia Partnership and Pathways Seventh Framework Programme

is gratefully acknowledged. The authors thank Alexander Mende and the people at RPM Risk & Portfolio

Management AB, Sweden for the many stimulating discussions and comments. The authors also thank the

participants at the Joint Conference on Institutional Investors/Hedge Funds and Emerging Market Finance,

17–18th September, 2015 at Ghent University, Belgium and the 13th INFINITI Conference on International

Finance, 8–9th June 2015 at Ljubljana, Slovenia. The views expressed in this paper are that of the author(s)

and do not necessarily reflect the views or opinions of RPM Risk & Portfolio Management AB.

JEL Classification: G11, G23

*Correspondence author, Department of Financial Economics, Ghent University, Sint-Pietersplein 5, Gent, Oost-

Vlaanderen, Belgium 9000. Tel: þ32-09-264-78-95, Fax: þ32-09-264-89-95, e-mail: gert.elaut@ugent.be

Received April 2015; Accepted November 2015

The Journal of Futures Markets, Vol. 36, No. 10, 992–1013 (2016)

© 2016 Wiley Periodicals, Inc.

Published online 4 February 2016 in Wiley Online Library (wileyonlinelibrary.com).

DOI: 10.1002/fut.21773

autocorrelations—exhibit distinctly different risk-return profiles and outperform funds that

exhibit lower degrees of serial correlation. A portfolio of more positively autocorrelated

Managed Futures funds displays higher risk-adjusted performance and lower drawdowns.

Application of multifactor models, including models using the recently proposed risk

factors suggested by Baltas and Kosowski (2012) as well as the more commonly used hedge

fund risk factors of Fung and Hsieh (2004), indicate a significantly positive risk-adjusted

excess return (“a”) of approximately 6% p.a. Interestingly, the models univocally suggest a

lower explanatory power in the case of the more positively serially correlated Managed

Futures funds. This finding of a low explanatory power of multifactor models coupled with

risk-adjusted outperformance corroborates other recent findings in the literature on

performance persistence in hedge funds.

1

In particular, Sun, Wang, and Zheng (2012) propose a “Strategy Distinctiveness Index”

(SDI) constructed as 1 minus the correlation between a hedge fund’s historical returns and

the returns of its peers. The aim of Sun et al. is to capture the degree to which hedge fund

managers follow unique investment strategies. The authors find that higher strategy

distinctiveness is associated with better future fund performance. Similarly, Titman and Tiu

(2011) show that hedge funds exhibiting lower R

2

swith regard to systematic factors have

higher Sharpe ratios, higher information ratios, and higher as. The authors conjecture that

funds that have more confidence in their abilities will expose their investors less to factor risk.

Our results are consistent with the above findings. Sorting Managed Futures funds on

the degree of serial correlation results in a subset of funds that outperform peers exhibiting

lower degrees of serial correlation. Coincidentally, these more positively serially correlated

funds’returns are found to be less well explained by existing multifactor models. This seems

to suggest that the serial correlation we observe is a consequence of the unique investment

strategies followed by these managers.

However, self-reported returns do not necessarily reflect all risks inherent to investing in

hedge funds and thus might overstate the actual return experience of investors. Therefore, we

explore several alternative explanations for the observed premium. Among others, we

consider attrition rates and the associated delisting bias as well as exposure to tail risk as

potential explanations for the observed outperformance. Despite slightly higher attrition

rates among more positively serially correlated managers, we find that a potential delisting

bias is unable to fully explain the observed outperformance.

The rest of this paper is structured as follows. The relevant literature is summarized and

discussed in section 2. Section 3 describes the Managed Futures space considered for the

analysis. In section 4, we outline the methodology used to determine the degree of

persistence in Managed Futures funds’self-reported returns. We analyze the risk-return

characteristics and potential drivers for the observed premium in section 5. Section 6

concludes.

2. RELATED LITERATURE

Evidence of performance persistence among hedge funds is, of course, not new. Although

early hedge fund literature gravitates towards a lack of performance persistence in hedge

funds’self-reported returns (see inter alia Brown, Goetzmann, & Ibbotson, 1999; Brown &

Goetzmann, 2003; Capocci & H€ubner, 2004; Malkiel & Saha, 2005), more recent

contributions present evidence of performance persistence.

In particular, Agarwal and Naik (2000) find persistence at the monthly frequency,

Baquero, ter Horst, and Verbeek (2005) find persistence at the quarterly level, and Agarwal,

1

We thank an anonymous referee for calling attention to this connection with the recent literature.

Risk-Return Characteristics of Managed Futures 993