Option trading after the opening bell and intraday stock return predictability

• Date: 01 September 2020
• Published date: 01 September 2020
• Author: Vijay Singal,Jitendra Tayal,Kelley Bergsma,Andy Fodor
• DOI: http://doi.org/10.1111/fima.12284

DOI: 10.1111/fima.12284

ORIGINAL ARTICLE

Option trading after the opening bell and intraday

stock return predictability

Kelley Bergsma1Andy Fodor1VijaySingal2Jitendra Tayal1

1College of Business, Ohio University, Athens,

Ohio

2PamplinCollege of Business, Virginia Tech,

Blacksburg, Virginia

Correspondence

JitendraTayal,College of Business, Ohio Univer-

sity,Copeland Hall, Athens, OH 45701.

Email:tayal@ohio.edu

Abstract

Prior literature finds that information is reflected in option markets

before stock markets, but no study has explored whether option

volume soon after market open has predictive power for intraday

stock returns. Using novel intraday signed option-to-stock volume

data, we find that a composite option trading score (OTS) in the

first 30 min of market open predicts stock returns during the rest

of the trading day. Such return predictability is greater for smaller

stocks, stocks with higher idiosyncratic volatility, and stocks with

higher bid–ask spreads relative to their options’ bid–ask spreads.

Moreover,OTS is a significantly stronger predictor of intraday stock

returns after overnight earnings announcements. The evidence sug-

gests that option trading in the 30 min after the opening bell has pre-

dictive power for intradaystock returns.

KEYWORDS

earnings announcements, intraday trading, option volume, stock

return predictability

1INTRODUCTION

Optiontrading is a critical component of efficient financial markets. It not only incorporates newly released information

into option prices but also signals the direction of subsequent stock price movements. The seminal study of Easley,

O’Hara, and Srinivas (1998) demonstrates how option volume predicts future stock returns through their model in

which informed investors choose to trade in equity or options markets. Subsequent empirical work at the daily and

weekly level confirms option tradingcontains information about future stock prices. For instance, Pan and Poteshman

(2006) document daily put–call ratios from trades opening new positions exhibit negativestock return predictability.

Similarly,Johnson and So (2012) report that the option-to-stock (O/S) volume ratio (Roll, Schwartz, & Subrahmanyam,

2010) is negatively related to future weekly stock returns. Ge, Lin, and Pearson (2016; hereafter GLP) disaggregate

the O/S ratio into eight signed measures, which also predict weekly stock returns. Although Chan, Chung, and Fong

(2002) find intradayoption volume has no power to predict stock returns, Hu (2014) documents that daily and intraday

option–induced order imbalances have significant stock return predictability.

c

2019 Financial Management Association International

Financial Management. 2020;49:769–804. wileyonlinelibrary.com/journal/fima 769

770 BERGSMA ETAL.

Increasingly, intraday stock return patterns are becoming the focus of a fast-growing literature. For instance,

Bogousslavsky (2016) explains intraday stocks returns’ negative autocorrelation as arising from infrequent portfolio

rebalancing. Gao, Han, Li, and Zhou (2018) document a new time-series return pattern in which the S&P 500 exchange-

traded fund’s (ETF’s) first half-hour return as measured from the prior day’s market close predicts the last half-hour

return. Their empirical evidence is consistent with Bogousslavsky’s (2016) model, as well as a model of trading near

market close on news released earlier in the day. Lou, Polk, and Skouras (2019) find the tug-of-war between retail

investors and institutional investorsdrives certain anomalies to accrue solely intraday or overnight.

In this paper, we contribute to both strands of literature on (a) option volumeand stock return predictability and

(b) the cross section of intradaystock returns. Specifically, we investigate whether a measure derived from signed O/S

ratios in the first 30 min after market open has predictivepower for stock returns during the rest of the day.1We uti-

lize a novelInternational Securities Exchange (ISE) dataset of intraday option volume from 2012 to 2014 for more than

2,000 firms. We construct a new firm-leveli ntradayoption trading score (OTS) that captures option traders’ bullish or

bearish sentiment regarding a particular stock. OTS is equal to bearish O/S ratios(call sales and put purchases) minus

bullish O/S ratios (call purchases and put sales). After controlling for stock-levelcharacteristics and short selling activ-

ity, we find that higher OTS during the first half hour of trading predicts significantly lower abnormal returns from

10:00 a.m. to 4:00 p.m.

Toexplore whether the predictive power of OTS stems from bullish or bearish option trades, we dissect OTS in the

first 30 min into its bullish and bearish components. Because Johnson and So (2012) emphasize that option trading

tends to reflect bad news more often than good news, one might expect bearish option trades to drive stock return

predictability. However, GLP report that purchases to open long call positions—a bullish trade—have the strongest

predictive power.We find that both the bullish and bearish sides of OTS in the first half hour significantly contribute to

stock return predictability for the rest of the day.

Next, we investigate a number of possible explanationsfor OTS’s predictive power, including liquidity, information

environment, short sale constraints, and overnight sentiment. Wemeasure absolute and relative liquidity using stock

bid–ask spreads (SBA), option bid–ask spreads (OBA),and the ratio between the two (SBA/OBA). We expect that when

options are more liquid relative to stocks—that is, SBA/OBA is high—then OTS will havea greater predictive power

because investors will prefer to trade in options before stocks to reduce their trading costs. In a tercile analysis, we

find that as SBA/OBA increases, the predictive power of OTSsignificantly increases. Moreover, our evidence suggests

that OTS has stronger stock return predictability for firms with smaller market capitalization and greater idiosyn-

cratic volatility.This result is consistent with the idea that small, volatile firms offer more profitable opportunities for

informed trading (Tookes,2008). Overall, we deduce that the predictive power of OTS is influenced by a number of

factors, including relative liquidity and information environment.

In further tests, we examine whether OTS’s predictive power is concentrated in a certain option moneyness cat-

egory or a particular option trade volume class (small/medium/largecustomer or firm). We find that OTS constructed

with out-of-the-money(OTM) or at-the-money (ATM) options demonstratesmore stock return predictability than OTS

constructed with in-the-money (ITM) options. These results support the notion that informed traders prefer options

with greater embedded leverage (OTMoptions) but sometimes trade off leverage for greater liquidity (ATM options),

corroborating Chakravarty,Gulen, and Mayhew (2004) and GLP.In addition, OTS generated using small trades by cus-

tomers only exhibits significant stock return predictability,whereas OTS generated using all other trade volume types

separately has insignificant coefficients. These findings are consistent with GLP who argue that order splitting is the

reason why small customer trades are themost informative.2

1Ourstudy complements the work of Hu (2014) and Patel, Putnins, Michayluk, and Foley (2018) who study intraday option-induced order imbalances and the

roleof options in stock price discovery, respectively, for more recent sample periods.

2GLP suggest that firm proprietary trades do not contain directional information because these traders may use complex spread and volatility option

strategies.

BERGSMA ETAL.771

Traditionalmodels on the information content of option volume for stock return predictability are based on private

information (e.g., Easleyet al., 1998; Pan & Poteshman, 2006). However, Engelberg, Reed, and Ringgenberg (2012) sug-

gest that informed traders’ advantage comes more from their ability to analyze—rather than anticipate—public infor-

mation. Cremers, Fodor,Murayev, and Weinbaum (2019) apply this concept to the options market. They find options

trading activity is abnormally high both immediately before news days and on news days,suggesting informed options

traders both anticipate news and process information effectively.Patel, Putnins, Michayluk, and Foley (2018) also sug-

gest options’ contribution to stock price discovery increases around information events. Numerous studies report

abnormal option volume prior to information events (e.g., Amin & Lee, 1997; Augustin, Brenner, & Subrahmanyam,

2015; Cao, Chen, & Griffin, 2005; Lung & Xu, 2014), but few studies except Cremers et al. (2019) examineoption vol-

ume on an information event day itself.

We study the predictive power of OTS immediately after overnight earnings announcements to examine how

traders process information.3Because many earnings announcements are released overnight (deHaan, Shevlin, &

Thornock, 2015; Lyle, Rigsby,Stephan, & Yohn, 2018), options markets may be an attractive venuefor trading on that

information in the first half hour after market open. In contrast, institutions may be less likely to trade in the stock

marketat that time due to wide bid–ask spreads, retail trader-driven overpricing, and higher volatility (Berkman, Koch,

Tuttle,& Zhang, 2012; Lou et al., 2019; McInish & Wood, 1992). Thus, we anticipate more information assimilation in

options markets relativeto the stock markets in the first 30 min after market open, given a greater relative proportion

of stock trading takesplace after 10:00 a.m. If option traders react more quickly to overnight earnings announcements

while stock traders respond more gradually,we expect OTS in the first half hour will predict stock returns during the

rest of the day to a greater extentfollowing overnight announcements.

Toexplore the information processing–based channel, we examine firms with overnight earnings announcements

as recorded in Institutional Broker’sEstimate System (I/B/E/S). We conduct a difference-in-difference analysis to com-

pare OTS on announcement versus nonannouncement days for these firms. The evidence suggests OTS has signifi-

cantly stronger stock return predictability after overnight earnings announcements, particularly for negative earnings

surprises. Our findings indicate OTShas a greater stock return predictability following information events, suggesting

a substantial number of investors tradingon overnight earnings announcements prefer option markets.

Last, we test the robustness of our intradayOTS measure by controlling for other option volume–based ratios, such

as the unsigned put–call ratio, Pan and Poteshman’s (2006) open buy put–call ratio,and the unsigned O/S ratio. Only

OTS and Panand Poteshman’s measure in the first half hour of trading have significant stock return predictability for

the rest of the day.Pan and Poteshman’s measure does not subsume the predictive power of OTS.

Overall, our results demonstrate intraday signed O/S ratioscontribute to stock return predictability. Our study is

part of the bridge connecting option to stock markets, built by Easleyet al. (1998), Hu (2014), Johnson and So (2012),

Pan and Poteshman (2006), Roll et al. (2010), and GLP, among others. Section 2 provides the motivation for our study.

Section 3 describes our data and methodology.Section 4 presents our empirical results. Section 5 concludes.

2MOTIVATION

The motivation for our study is from prior literature on the relation between option volume and stock prices. Black

(1975) argues that many informed traders will prefer option markets over stock markets because short-term trad-

ing costs are lower and potential gains are higher for options. In Easley et al.’s (1998) multimarket equilibrium model,

informed investors choose to trade in option or equity marketsbased on their private information and prefer options

in certain cases based on these instruments’ inherent leverage and liquidity.Pan and Poteshman (2006) further con-

nect prior information-based models to informed options trading. Using proprietary Chicago Board Options Exchange

3Blauand Wade (2012) suggest some options trading prior to announcements is speculative rather than informed. We avoidthis issue altogether by examining

onlyoption trading after announcements.