How about selling commodity futures losers?

• Date: 01 December 2019
• Author: Kyung Yoon Kwon,Jangkoo Kang
• DOI: http://doi.org/10.1002/fut.22051
• Published date: 01 December 2019

J Futures Markets. 2019;39:1489–1514. wileyonlinelibrary.com/journal/fut © 2019 Wiley Periodicals, Inc.

|

1489

Received: 2 May 2019

|

Accepted: 11 August 2019

DOI: 10.1002/fut.22051

RESEARCH ARTICLE

How about selling commodity futures losers?

Jangkoo Kang

1

|

Kyung Yoon Kwon

2

1

College of Business, Korea Advanced

Institute of Science and Technology,

Dongdaemoon‐gu, Seoul, South Korea

2

Department of Accounting and Finance,

Strathclyde Business School, University of

Strathclyde, Glasgow, UK

Correspondence

Kyung Yoon Kwon, Department of

Accounting and Finance, Strathclyde

Business School, University of

Strathclyde, 199 Cathedral Street,

Glasgow G4 0QU, UK.

Email: arari1115@gmail.com

Funding information

National Research Foundation of Korea,

Grant/Award Number: NRF‐

2017S1A5B5A07064108

Abstract

This paper explores the benefits of extending the investment universe to

commodity futures, from the perspective of momentum traders. We find that

the growth‐optimal portfolio includes negative (positive) weights on commodity

futures losers (stock winners). Motivated by this finding, we construct a joint

momentum strategy, buying stock winners and selling commodity futures

losers, and show that it generates an average monthly return of up to 1.91% and

provides much lower skewness (0.04) and kurtosis (1.27) than a traditional

stock momentum strategy. It also greatly improves profitability, especially in

unfavorable market states, and thus effectively manages tail risk.

KEYWORDS

commodity futures, market anomalies, momentum, momentum crashes, tail risk

JEL CLASSIFICATION

G10; G11; G12

1

|

INTRODUCTION

Momentum is one of the most studied phenomena in the asset pricing literature. Momentum indicates that assets with

recent outperformance (underperformance) relative to others tend to continuously outperform (underperform) in the

future. The stock momentum strategy, which buys recent stock winners and sells recent stock losers, has been shown to

be attractive, generating positive and significant returns, on average. In contrast, this strategy exhibits great negative

skewness and high kurtosis. Recent empirical studies have reported that the stock momentum strategy occasionally

generates large negative returns that are especially notable in bear markets or during market rebounds (Barroso &

Santa‐Clara, 2015; Daniel & Moskowitz, 2016; Daniel, Jagannathan, & Kim, 2012; Grundy & Martin, 2001). Specifically,

Barroso and Santa‐Clara (2015) report that the stock momentum strategy delivered a return of −91.59% over 2 months

in 1932 and a return of −73.42% over 3 months in 2009, both when the stock market was rebounding from a severe

recession.

Momentum profit was first documented in the U.S. stock market by Jegadeesh and Titman (1993); however, the

literature has since accumulated evidence of the momentum phenomenon in other markets, including the commodity

futures market (Asness, Moskowitz, & Pedersen, 2013; Erb & Harvey, 2006; Kang & Kwon, 2017; Lutzenberger, Gleich,

Mayer, Stepanek, & Rathgeber, 2017; Miffre & Rallis, 2007). In this paper, we examine a momentum strategy in the

investment universe consisting of the commodity futures markets as well as the stock market. Investments in

commodity futures markets have increased rapidly over the last few years.

1

Commodity futures are one of the most

important hedge instruments and are also useful in enhancing diversification benefits in the asset universe due to their

1

According to index investment data provided by the U.S. Commodity Futures Trading Commission, the total value of index investments in the U.S. commodity futures market was about $119billion

at the end of 2008 but grew to about $187 billion by the end of 2014. At the end of 2010, when the financial markets were affected by the global financial crisis, the total value of index investments in

the US commodity futures market was especially high, at $293 billion.

relatively low correlation with other assets. The academic literature has extensively investigated hedging effects and the

diversification benefits of commodity futures (Belousova & Dorfleitner, 2012; Conover, Jensen, Johnson, & Mercer,

2010; Daskalaki & Skiadopoulos, 2011), but the benefits of extending the investment universe to commodity futures

from the perspective of momentum traders have not yet been explored. We expect momentum profits using commodity

futures and stocks together to remain significantly positive but with less risk, because commodity futures momentum

returns have a very low correlation with stock momentum returns. Kang and Kwon (2017) document a significant

positive relation between the two momentums, using international commodity futures market data, but report that this

correlation is quite low and that the commodity futures momentum cannot be fully explained by the stock momentum.

Our results also show that the correlation between the stock momentum and the commodity futures momentum ranges

from −0.035 to 0.037, which is quite low. This low correlation cannot be obtained in the case of a momentum strategy in

the other markets. Specifically, during our sample period, we find that our stock‐only momentum has correlations of

0.35, 0.15, and 0.24, respectively, with the equity index, currency, and fixed income momentums constructed by Asness

et al. (2013).

2

This paper examines how we can improve the performance of the stock market momentum strategy by combining it

with the commodity futures momentum strategy in terms of risk‐adjusted returns, utility gains, and alleviating tail risk.

Using a data set from January 1979 to June 2015, we first investigate a way to combine the two momentums by

estimating the growth‐optimal portfolio. One way to expand the investment universe into commodity futures in

addition to the stock market from the momentum strategy is to create a momentum strategy that treats all assets as one

asset class and selecting the winners and loser among all the assets. However, commodity futures are regarded as a

distinctive asset class compared to stocks having different nature and also different return distributions. Thus, treating

them as one asset class would not be an appropriate approach. Thus, we take into consideration only the combinations

of the stock winner portfolio, the stock loser portfolio, the commodity futures winner portfolio, and the commodity

futures loser portfolio. It is rare in the literature for a momentum strategy to be constructed across markets, such as the

stock market and the commodity futures market. Though Asness et al. (2013) examine the profitability of momentum

strategies in diverse asset universes, they do not try to construct a cross‐market momentum strategy that outperforms

single‐market momentum strategies.

As a guide to construct a “better”momentum strategy, we examine the growth‐optimal momentum strategy that is a

combination of the four stock market and commodity futures market winners and losers and other stock portfolios,

such as size, value, profitability, and investment portfolios. Because the growth‐optimal portfolio is the optimal portfolio

for investors with a log‐utility function, it is the optimal portfolio for at least some investors. Our empirical estimates

show that (a) both stock losers and winners have large positive weights but stock winners carry a much larger positive

weight than stock losers in the growth‐optimal portfolio, (b) commodity futures winners have a positive weight but a

much smaller weight than stock winners, and (c) commodity futures losers have a negative weight in the growth‐

optimal portfolio. The growth‐optimal portfolio shows that a larger weight for winners than for losers in a single

market, such as only the stock market or only the commodity futures market, is optimal, justifying the single‐market

momentum strategies. However, the results suggest another and probably stronger momentum strategy consisting of a

long position in stock winners and a short position in commodity futures losers. This is the momentum strategy that we

examine in detail in this paper.

Selling commodity futures losers has its advantages. First, commodity futures losers have much lower skewness and

kurtosis relative to stock losers, which results in the lower tail risk of the commodity futures momentum strategy. For

example, Daniel and Moskowitz (2016) report that the skewness values of the momentum return in global equity

markets, fixed income markets, and foreign exchange markets are −0.34, −0.24, and −0.54, respectively, but commodity

futures momentum returns have a skewness of only 0.01 (see table 11 in Daniel & Moskowitz, 2016). Our results also

show that short‐term commodity futures momentum returns have a skewness of 0.160 and a kurtosis

3

of 0.571. This

favorable feature of the return distribution of the commodity futures momentum is known in the literature but has not

attracted much attention. Daniel and Moskowitz (2016) document that the stock momentum strategy is exposed to

crash risk and stock momentum crashes are attributable to the short side of the momentum strategy, that is, stock

losers. Thus, we expect that the substitution of commodity futures losers in the place of stock losers alleviates the tail

risk of the traditional stock momentum strategy.

2

See https://www.aqr.com/Insights/Datasets.

3

In this paper, kurtosis indicates excess kurtosis, and so the normal distribution has a value of 0.

1490

|

KANG AND KWON