ODD LOT ORDER AGGRESSIVENESS AND STEALTH TRADING

• DOI: http://doi.org/10.1111/jfir.12123
• Published date: 01 June 2017
• Author: Brian Roseman,Hardy Johnson
• Date: 01 June 2017

ODD LOT ORDER AGGRESSIVENESS AND STEALTH TRADING

Hardy Johnson

Kansas State University

Brian Roseman

California State University, Fullerton

Abstract

We investigate the degree to which orders are aggressively priced, paying particular

attention to odd lot orders, and examine whether odd lot orders are being successfully

used in stealth trading strategies. We find that odd lot orders execute at higher

frequencies than larger orders, which is due to odd lot orders being aggressively priced.

We find that microstructure changes have not altered the nature of price aggressiveness,

but its determinants hold different explanatory power for odd lot orders. We find

evidence that informed traders are shredding their orders into odd lot orders and stealth

trading is permeating odd lot denominations.

JEL Classification: G10, G14

I. Introduction

Market participants make two joint decisions when submitting orders to equity

exchanges. The first is the size of the order they wish to submit. Traders possess sets of

information and preferences that dictate the size of the position they want to assume.

However, financial markets are sensitive to large order submissions, encouraging

informed traders to split up their position into smaller order sizes to minimize the market

impact of their order submission. Large orders are generally believed to be more

informed, while small orders are believed to contain less information.

1

The second (and

equally important) decision that traders make is the price of their order. The order price is

a major determinant of whether the order will execute and a determinant of the market’s

reaction to the order submission. The probability of execution, and the resulting market

reaction, must be weighed against the trader’s need for a preferable price and liquidity.

With regard to price, the trader can choose to place orders aggressively, which helps

ensure execution but may influence the market price of the security. Conversely, orders

The authors thank for their helpful comments and feedback: Chairat Chuwonganant, Ryan Davis, Todd

Griffith, conference participants at the 2016 Financial Management Association meeting in Las Vegas, NV, and the

2016 Southern Finance Association meeting in Sandestin, FL, and the editors, associate editors, and reviewers at

the Journal of Financial Research.

1

Barclay and Warner (1993) and Alexander and Peterson (2007) find evidence of informed trading in trade

sizes between 500 and 9,900 shares. Studies such as Dyl and Maberly (1992) and Lakonishok and Maberly (1990)

use small, odd lot trade sizes as a proxy for uninformed investors. Additionally, Upson and Johnson (2017) do not

find evidence that odd lot trades are more informed than larger trade sizes.

The Journal of Financial Research Vol. XL, No. 2 Pages 249–281 Summer 2017

249

© 2017 The Southern Finance Association and the Southwestern Finance Association

RAWLS COLLEGE OF BUSINESS, TEXAS TECH UNIVERSITY

PUBLISHED FOR THE SOUTHERN AND SOUTHWESTERN

FINANCE ASSOCIATIONS BY WILEY-BLACKWELL PUBLISHING

can be submitted passively, where the probability of execution depends on the market

price moving toward the order price.

With the advent of algorithmic trading, the human capital needed to facilitate

sophisticated and complex strategies has significantly decreased. For example,

computers make it easy for traders that want to assume large positions to lessen the

market impact by shredding their orders into smaller denominations. This practice has

been well documented in the stealth trading literature. The computational overhead

required for the assumption of large positions has decreased sufficiently that even smaller

orders, including odd lot orders (orders for less than 100 shares) may be used to fill a

position. Additionally, a second strategic reason traders may employ odd lots and

computers is to explore the nature of the market reactions to order submissions. In

the model of Clark-Joseph (2015), sophisticated traders submit small and aggressive

orders to observe the market’s reaction, a practice termed “exploratory trading.”

The information gained allows the trader to make informed decisions about future order

submission.

Our study focuses on the aggressiveness with which orders are submitted, paying

particular attention to small, odd lot orders.

2

We do so to three ends. First, we examine

the pricing of odd lot orders to determine whether they are being used to trade

aggressively to a greater extent than their larger sized counterparts, providing evidence

that stealth trading has entered odd lot denominations. Second, we look for evidence of

stealth trading among odd lot orders by assessing the market impact and profitability of

aggressively priced, odd lot orders through effective spreads and five-minute trading

profits. Third, we revisit the determinants of order aggressiveness and assess whether

market microstructure changes over the last several decades have influenced the nature of

order aggressiveness. We find that odd lot orders are priced more aggressively and the

most aggressive odd lot orders are informed orders. We show that the most aggressive

odd lot orders, which we contend are stealth traders, are successfully decreasing their

market impact by using odd lot denominations. We conclude the nature of price

aggressiveness has not changed since prior studies, but the determinants of odd lot order

pricing are very different. We propose the differences are due, at least in part, to traders

mitigating their exposure to price risk.

Until recently, odd lot order s and trades have generally bee n ignored in the

capital markets literature. Odd lots were assumed to be of little importance in equity

markets. The prevailing theory was that large, informed institutions trade in much

larger blocks than 100 shares , whereas uninformed retai l traders were investing

through odd lot denominations bec ause of the economies of scale necessar y to trade

larger order sizes (see Ritter 1988; Dyl and Maberly 1992; Lakonishok and Maberly

1990). Our preliminary resul ts show that odd lot orders have much higher executi on

rates than 100þshare orders. Hi gher execution rates could ce rtainly be caused by odd

2

Although odd lot trades are treated the same as any other trade size classification in our sample period,

collectively they form a structural break that makes them easy to distinguish from other trade sizes. Motivated by

theoretical work showing that informed traders strategically choose order submission strategies (as in Kyle 1985;

Admati and Pfleiderer 1988), odd lots represent the smallest size denomination into which larger orders can be

broken.

250 The Journal of Financial Research

lot orders being more aggressi vely priced than 100þshare orders, and we find that this

is the case.

O’Hara, Yao, and Ye (2014) and Johnson, Van Ness, and Van Ness (2015) find

that there is information contained within odd lot trades, as odd lot trades contribute to

price discovery to a greater extent than their proportion of volume. However, Upson and

Johnson (2017) find that odd lot orders are at least not more informed than 100þshare

orders, and most likely are uninformed orders. We do not contend that all odd lot orders,

or even a preponderance of odd lot orders, are submitted by informed traders. But we do

find odd lot orders that aggressively cross the limit order book and trade with immediacy

have positive and economically significant trading profits.

The rationale that informed traders strategically place their orders to decrease the

market impact is straightforward. In the model of Kyle (1985), investors who possess

private information regarding an asset’s price will strategically separate large orders into

smaller order sizes spread across the trading day, allowing them to extract profits from

market makers. In the model of Admati and Pfleiderer (1988), informed traders choose to

strategically place informed orders during periods of high volume to avoid alerting other

market participants. Both models highlight the sensitivity of markets to large and

informed trades, requiring investors trading large volumes to choose smaller order sizes.

The stealth trading literature (i.e., Barclay and Warner 1993; Chakravarty 2001;

Alexander and Peterson 2007) provides evidence that traders break up large order sizes

into multiple small and medium order sizes to reduce the information impact.

The existing empirical evidence provides baseline predictions regarding the

optimal order sizes employed by traders attempting to hide their information. Early

evidence is provided by Barclay and Warner (1993), who use a sample of 108 tender

offers from 1981 to 1984. From their results, they find that medium size trades, which are

classified as trades of 500–9,999 shares, have large and significant price impacts that are

disproportionately larger than the amount of volume traded. Furthermore, Hasbrouck

(1995) and Chakravarty (2001) confirm that medium size trades contain a

disproportionate amount of information and add that institutional traders employing

medium size trades have larger price impacts than individual traders. Although traders

certainly have always had the ability to split large orders into smaller order sizes, the

basis for using medium trades during the periods of earlier studies is that large orders

draw the attention of other market participants, whereas order sizes that are too small

require additional trading costs and commissions, as well as an increase in cognitive

overhead (Berkowitz, Logue, and Noser 1988; Chakravarty 2001; Alexander and

Peterson 2007). As a result, medium size orders were likely the optimal size to conduct

stealth trading before electronic and algorithmic trading.

As markets have shifted toward electronic trading, the strategies employed by

traders have changed as well. As discussed in Hendershott, Jones, and Menkveld (2011),

before the widespread use of electronic markets, large trades submitted by institutions

would be split into smaller orders and discretely traded on the trading floor of the New

York Stock Exchange (NYSE) in an effort to avoid moving the security price. With the

advancement of electronic trading, such as the NYSE Autoquote system introduced in

2003, computer algorithms are able to replicate the job of a floor trader efficiently and

automatically. A large “parent”order can be sliced into multiple “child”orders with

Odd Lot Order Aggressiveness 251