Predicting opening spot prices using extended futures trading

• Date: 01 April 2019
• DOI: http://doi.org/10.1002/for.2554
• Author: Janchung Wang,Sunwu Winfred Chen,Bo‐Ting Wang
• Published date: 01 April 2019

RESEARCH ARTICLE

Predicting opening spot prices using extended futures

trading

Janchung Wang

1

| Sunwu Winfred Chen

2

|Bo‐Ting Wang

3

1

Department of Money and Banking,

National Kaohsiung University of Science

and Technology, Kaohsiung, Taiwan

2

Department of International Business

Management, Shih Chien University

Kaohsiung Campus, Kaohsiung, Taiwan

3

Department of Accounting, National

Chengchi University, Taipei, Taiwan

Correspondence

Janchung Wang, Department of Money

and Banking, National Kaohsiung

University of Science and Technology, 1

University Road, Yanchao, Kaohsiung,

Taiwan, ROC.

Email: janchung@nkust.edu.tw

Funding information

Ministry of Science and Technology,

Taiwan, Grant/Award Number: NSC 101‐

2410‐H‐327‐027

Abstract

Previous studies found that extended futures trading contains useful informa-

tion in explaining subsequent overnight spot returns. This study therefore com-

pares the performance of using the extended trading of the TAIFEX (Taiwan

Futures Exchange) index futures and single‐stock futures to predict their open-

ing underlying spot prices. Furthermore, according to the efficient market

hypothesis, the share price fully reflects all the information available and

should adjust to new information instantaneously. However, several studies

have demonstrated that short‐sales restrictions delay the speed of price adjust-

ment to negative information. The relevant question is whether short‐selling

restrictions also slow down the speed at which the opening spot price adjusts

to the new information revealed through extended futures trading, and thus

reducing the price prediction function of extended futures trading. The empir-

ical results find that using the opening futures price and the prediction method

proposed in this study can more accurately predict the opening spot price on

the same day. Furthermore, the performance of using the extended trading of

index futures to predict the opening spot index price is superior to that of using

the extended trading of single‐stock futures to predict the opening stock price.

Finally, as found in previous studies, short‐selling restrictions also slow down

the speed of stock price adjustment to the new information revealed through

extended futures trading. Thus both the up‐tick rule and the short‐selling bans

(especially the latter) negatively affect the price forecasting performance of

extended futures trading.

KEYWORDS

cost of carry model, extended futures trading, opening spot price, rolling regression, short‐selling

restrictions

1|INTRODUCTION

In perfect capital markets, the index futures price and its

underlying spot index simultaneously reflect new infor-

mation, and there is no lead–lag relationship between

the two prices. However, in real imperfect capital mar-

kets, owing to several factors, including transaction costs,

short‐sale restrictions, leverage effect, order delays,

liquidity, and nonsynchronous trading, futures markets

and their underlying spot markets do not reflect new

information simultaneously, which creates a lead–lag

relationship between the two prices. Generally speaking,

informed traders prefer futures markets over their under-

lying spot markets. As a result, literatures regarding price

Received: 4 May 2018 Revised: 17 August 2018 Accepted: 25 August 2018

DOI: 10.1002/for.2554

Journal of Forecasting. 2019;38:155–174. © 2018 John Wiley & Sons, Ltd.wileyonlinelibrary.com/journal/for 155

discovery of futures markets are more inclined to support

that the index futures price should lead the underlying

spot index, and that the index futures price can more effi-

ciently convey useful information to investors than the

underlying spot index (e.g., Kawaller, Koch, & Koch,

1987; Kim, Szakmary, & Schwarz, 1999; Puttonen, 1993;

Stoll & Whaley, 1990). Additionally, the unbiased expec-

tations hypothesis implies that the current futures price

contains all relevant and available information regarding

the future spot price. Black (1986) and Grossman (1989)

have demonstrated that futures markets provide

informed traders with all relevant information regarding

the future spot price.

Additionally, futures markets cannot only instanta-

neously reflect new information but also convey relevant

information to spot markets, where information is rela-

tively insufficient. A few studies have found that the index

futures trading before the opening and after the closing of

the stock markets (briefly called the extended futures trad-

ing) contains information content about subsequent spot

index returns, and can be used to predict subsequent open-

ing spot prices. For example, according to the concept of

nonsynchronous trading mentioned by Fisher (1966) and

Scholes and Williams (1977), and the view presented by

Grossman (1977, 1989) that futures markets act as a con-

duit for transmitting useful information from informed

traders to uninformed traders, the empirical results of

Hiraki, Maberly, and Takezawa (1995) indicated that the

unexpected returns of the Nikkei 225 index futures for

the period between the closing time of the spot market

(3:00 pm) and the closing time of the futures market

(3:15 pm) (briefly called the post‐close extended session)

contain useful information of informed traders and can

be used to predict subsequent overnight spot returns.

Moreover, using the Hang Seng index futures, the empiri-

cal findings of Cheng, Jiang, and Ng (2004) revealed that

the unexpected futures returns for the period between the

opening of the index futures market (9:45 am) and the

opening of the stock market (10:00 am) (briefly called the

pre‐open extended session) contain useful information in

explaining subsequent overnight spot returns.

From the above analysis, extended futures trading

contains information content regarding subsequent over-

night spot returns, which can be used to predict subse-

quent opening spot prices. The Taiwan Futures

Exchange (TAIFEX) launched the Taiwan Stock

Exchange capitalization weighted stock index futures

(TAIEX futures), the Taipei Exchange capitalization

weighted stock index futures (OTC index futures), and

single‐stock futures (SSFs) contracts on July 21, 1998,

October 8, 2007, and January 25, 2010, respectively. The

trading hours of all TAIFEX futures contracts run from

8:45 am to 1:45 pm during the sample period. However,

the regular trading hours for stocks listed on the Taiwan

Stock Exchange and stocks traded on the Taipei Exchange

are from 9:00 am to 1:30 pm. In other words, the futures

market opens earlier and closes later than the stock mar-

kets. We therefore use in‐sample and out‐of‐sample suc-

cess rates and absolute percentage forecasting errors to

examine the performance of using the extended trading

of the TAIEX futures, OTC index futures, and SSFs to pre-

dict their opening underlying spot prices.

The efficient market hypothesis postulates that stock

prices fully reflect all available information and react

immediately to new information. However, numerous

studies have demonstrated that short‐selling restrictions

prevent investors from responding to bad news and slow

the downward adjustment of shares prices to negative

information. Diamond and Verrecchia (1987) concluded,

based on their theoretical model, that short‐selling bans

lead to asymmetric price discovery and information dis-

semination processes, which slows down the speed of

stock price adjustment to private and public information

(especially to bad news). Beber and Pagano (2013) inves-

tigated the effect of short‐selling bans on stock markets

in 30 different countries and found that the short‐selling

bans imposed during the 2007–2009 global financial crisis

reduce the speed of price discovery, and more so in bear

markets. Puttonen (1993) explored the spot–futures

dynamics in Spain and observed that index futures prices

lead spot index prices. Puttonen further found that short‐

selling restrictions reduce the speed of price adjustment

in the stock market. Fung and Draper (1999) divided

the study period into three subperiods based on restric-

tions imposed on short selling to examine the impact of

short‐selling constraints on the Hang Seng index futures

mispricing. Fung and Draper found that relaxing short‐

selling restrictions reduces the degree and frequency of

mispricing and increases market efficiency. These results

suggest that the lifting of short‐selling constraints can

increase the speed of stock price adjustment to negative

information. Fung and Jiang (1999) used an error correc-

tion model, whereas Jiang, Fung, and Cheng (2001)

employed a Sims‐style Granger causality test to examine

the effect of short‐selling restrictions on the lead–lag rela-

tionship between the Hang Seng index futures and the

underlying spot index. The results indicated that relaxing

short‐selling restrictions can increase the speed of stock

price adjustment to negative information, which

strengthens the contemporaneous relationship between

the index futures and the underlying spot index. Bris,

Goetzmann, and Zhu (2007) compared stock price adjust-

ments with and without short‐selling restrictions in dual

listing markets. They found slower price adjustment to

negative information under more stringent short‐selling

restrictions.

156 WANG ET AL.