Intraday time‐series momentum: Evidence from China

• Author: Muzhao Jin,Youwei Li,Fearghal Kearney,Yung Chiang Yang
• Date: 01 April 2020
• Published date: 01 April 2020
• DOI: http://doi.org/10.1002/fut.22084

J Futures Markets. 2020;40:632–650.wileyonlinelibrary.com/journal/fut632

|

© 2019 Wiley Periodicals, Inc.

Received: 22 September 2017

|

Accepted: 27 November 2019

DOI: 10.1002/fut.22084

RESEARCH ARTICLE

Intraday time‐series momentum: Evidence from China

Muzhao Jin

1

|

Fearghal Kearney

1

|

Youwei Li

2

|

Yung Chiang Yang

3

1

Queen’s Management School, Queen’s

University Belfast, Belfast, UK

2

Hull University Business School,

University of Hull, Hull, UK

3

UCD College of Business, University

College Dublin, Dublin, Ireland

Correspondence

Youwei Li, Hull University Business

School, University of Hull, Hull HU6 7RX,

UK.

Email: youwei.li@hull.ac.uk

Funding information

National Natural Science Foundation of

China, Grant/Award Number: 71571197

Abstract

This study conducts an investigation of intraday time‐series momentum across

four Chinese commodity futures contracts: copper, steel, soybean, and soybean

meal. Our results indicate that the first half‐hour return positively predicts the

last half‐hour return across all four futures. Furthermore, in metals markets, we

find that first trading sessions with high volume or volatility are associated with

the strongest intraday time‐series momentum dynamics. Based on this, we

propose an intraday momentum informed trading strategy that earns a return in

excess of standard always long and buy‐and‐hold benchmarks.

KEYWORDS

intraday predictability, momentum, time‐series

JEL CLASSIFICATION

G12; G13; G15

1

|

INTRODUCTION

Momentum effect can be broadly defined as the empirically observed tendency for rising asset prices to continue rising

and falling prices to continue falling (Jegadeesh & Titman, 2001). There are two main strands in the momentum

literature, cross‐sectional and time‐series. Cross‐sectional momentum uses a security’s past outperformance relative to

its peers to predict future outperformance (Jegadeesh & Titman, 1993). In contrast, time‐series momentum uses a

security’s own past return to predict its future return (Moskowitz, Ooi, & Pedersen, 2012). Focusing on the more

recently proposed time‐series momentum dynamic, we contribute by analyzing the intraday momentum effect in the

increasingly important and as yet unstudied Chinese commodity futures market.

The four largest metals and agricultural futures markets in the world are located in China. Motivated by its large

size, we employ a unique intraday data set of Chinese commodities. Our work contributes to the literature by closely

analyzing intraday time‐series momentum properties of the four futures: Copper, Steel Rebar, Soybean, and Soybean

Meal. We seek to answer three simple questions: (a) Is an intraday time‐series momentum pattern present, whereby the

return in the opening period predicts the return in the market closing period for Chinese commodities? (b) Can such a

pattern be economically exploited through the implementation of a trading strategy? (c) What explanations are there for

the presence of such patterns?

Time‐series momentum focuses on endogenously achieving excess returns using an asset’s own return series. It is

broadly similar to testing the random walk hypothesis for a single asset or examining the persistence of an asset’s

returns. Time‐series momentum has widespread applicability, as evidence has been documented for numerous asset

classes around the world, including those subject to short sale constraints. Commodity trading advisors (CTAs)

achieved strong performance using time‐series momentum strategies, in particular, during the 2008 global financial

crisis (Baltas & Kosowski, 2013), piquing the interest of market practitioners and academic scholars, alike. This provides

strong motivation, from both an academic and market practitioner perspective, to study the intraday time‐series

momentum effect in commodity markets.

Most existing cross‐sectional and time‐series momentum studies are based on interday trading. However, driven by

technological developments and enhanced data availability in recent years, intraday high‐frequency momentum

strategies, such as those that we consider, can now be examined. For instance, Kang (2005) uses two thousand NYSE

stocks, documenting an intraday cross‐sectional momentum strategy that generates winners and losers according to

their hourly returns, and compares their performance over a given trading day. The results suggest that winners

outperform losers for about one and a half hours, with a subsequent reversal observed for large stocks. Small stocks, in

contrast, tend to continue their momentum effect throughout the day. Venter (2009) also follows the winner/loser

ranking method to study intraday cross‐sectional momentum on Johannesburg Stock Exchange stocks, arguing that a

cross‐sectional momentum effect can be uncovered within a day. Furthermore, Gao, Han, Zhengzi Li, and Zhou (2018)

and Sun, Najand, and Shen (2016) utilize the time‐series momentum approach of Moskowitz et al. (2012) to study the

S&P 500, they document an intraday time‐series momentum pattern that the first half‐hour return is able to predict the

last half‐hour return. Similarly, Elaut, Kevin, and Michael (2018) find that the first half‐hour return can be used to

predict the last half‐hour return in the RUB‐USD FX market. Liu and Tse (2017) focus on S&P 500 ETFs as well as 12

international index futures contracts, finding that, despite the first half‐hour return not exhibiting predictive power,

that overnight returns positively predict the last half‐hour return. As outlined, the majority of both intraday cross‐

sectional and time‐series momentum studies are based on stocks or stock indices, with intraday momentum in

commodity markets remaining an open research question.

We present evidence consistent with Gao et al. (2018), with the first half‐hour return significantly predicting the last

half‐hour return across all four commodity futures markets. We provide two explanations for this intraday momentum.

The first explanation is the liquidity provision of intraday traders. For instance, day traders provide liquidity by taking

the opposing direction to the rest of the market, however, these day traders subsequently close their positions at the end

of the day to avoid overnight risk. Since trades occur more rapidly at the beginning and end of a given trading day

(Hora, 2006), many of the day traders provide liquidity in the morning and subsequently close out their positions at the

end of the day, causing prices during the last trading session to move in the same direction as the first. The second

explanation is the trading time preference of the strategically informed traders. The informed traders prefer to trade

during high trading volume periods to camouflage their information and limit their price impact. Therefore, the

informed traders primarily execute their transactions at the beginning and near the end of the day because of the

U‐shape intraday trading volume pattern (Hora, 2006).

We also find that for metals futures, the opening trading sessions with the highest volume or volatility have the

greatest intraday time‐series momentum predictability. We confirm that this dynamic also holds when specifying a

15‐min time‐frequency. However, we did not find the above phenomenon for agricultural futures, with one possible

explanation being the high proportion of noise traders active in the market. Furthermore, we also find, that using

intraday time‐series momentum as a trading strategy earns greater abnormal profits relative to “always long”and “buy‐

and‐hold”strategies.

The rest of the paper is organized as follows: Section 2 is devoted to the description of the data used in our study. In

Section 3, we present the intraday time‐series momentum methodology employed in the paper. We present our main

findings in Section 4, Section 5 assess the robustness of our results, with Section 6 concluding.

2

|

DATA DESCRIPTION

2.1

|

Commodity markets in China

Using the 2014 Futures Industry Association (FIA) world ranking in Table A1 of Appendix A (Acworth, 2015), we

observe that trading volume in Chinese markets occupies the top four positions worldwide, for both metals and

agricultural contracts in 2014. This paper focuses on four Chinese commodity futures contracts: copper, steel rebar,

soybean, and soybean meal.

Copper futures are the longest established metals market in China. The market was established in China in 1992,

however, it was not until rectification in 2003, that it became a relatively mature market. Again based on the FIA

2014 report (Acworth, 2015), the trading volume of Chinese copper contracts means it is the largest copper futures

market in the world and fourth in the rankings of the top metals futures and options contracts. As well as copper,

China is also the largest steel producer and consumer in the world, based on the FIA report (Acworth, 2015). This is

despite the market for steel rebar in China having only been established in 2009. The trading volume of Chinese

JIN ET AL.

|

633