CAN NONLOCAL TRADERS CAPTURE THE LOCAL INFORMATION ADVANTAGE AND PROFIT?

• Published date: 01 March 2019
• Author: Drew B. Winters,Artem Meshcheryakov
• DOI: http://doi.org/10.1111/jfir.12175
• Date: 01 March 2019

CAN NONLOCAL TRADERS CAPTURE THE LOCAL INFORMATION

ADVANTAGE AND PROFIT?

Artem Meshcheryakov

San Jose State University

Drew B. Winters

Texas Tech University

Abstract

Market makers located in geographic proximity (local) to companies possess a local

information advantage that comes from access to soft information. We study whether a

nonlocal trader can capture the local information advantage and profit without

relocating. We develop a trading strategy for the nonlocal trader that generates “buy”

and “sell”signals for stocks based on quotes of local market makers. Our findings

suggest it is possible, albeit difficult, for nonlocal traders to extract local information

from local market makers’quotes. Using limit orders from buy signals, we generate up

to 7.6 basis points of abnormal return per day.

JEL Classification: G14

I. Introduction

Semistrong efficient markets allow traders with an information advantage to profit from

their information. With the advances of modern digital technology, the finance literature

has moved to divide information into two categories: hard (easy to digitize and transmit)

and soft (difficult to digitize and transmit) (Petersen 2004). Hard information, such as

financial statements or analysts’reports, are easy to digitize and transfer any distance

without deterioration of quality and value, making the location of market participants less

important. Soft information is difficult to digitize and does not transmit well, arguably

providing an information advantage for local traders. We explore whether it is possible

for nonlocals to access local information and profit.

The market microstructure literature demonstrates that local market makers

enjoy multiple benefits from being located near companies in which they make a market.

1

These benefits make local market makers more effective in market-making decisions,

more profitable in trading local stocks, and overall more informed (Kedia and Zhou 2011;

Anand et al. 2011). Therefore, we assume that the local information advantage is

reflected in local market makers’quotes.

1

Schultz (2003) suggests that market makers’inventory risk is reduced by the constant and predictable order

flow from local clients and that the adverse selection risk of trading against informed traders is lower for locals than

for nonlocals.

The Journal of Financial Research Vol. XLII, No. 1 Pages 41–69 Spring 2019

DOI: 10.1111/jfir.12175

41

© 2019 The Southern Finance Association and the Southwestern Finance Association

Open electronic limit order book (LOB) markets, such as NASDAQ, allow

traders to observe market makers’quotes.

2

Using NASDAQ Historical ITCH trading

data for the first six months of 2013,

3

we study whether it is possible for nonlocal traders

to capture the local information advantage and profit without relocating by observing

local market makers’quotes. In other words, we study whether geographic proximity

(market maker is located in the same metropolitan statistical area [MSA] as company

headquarters) is a necessary condition to benefit from soft information using

microstructure data.

For the purpose of this study, we develop a trading algorithm that generates buy

and sell signals for a nonlocal trader based on quotes of local market makers. We find that

our generated buy signals accurately predict the direction of short-term stock price

changes between 53% and 63% of the time.

4

At the same time, sell signals are not

informative in any tested month. These results suggest that nonlocal traders can extract

local information from local market makers’quotes.

Finding information in buy signals only, we test whether nonlocal traders can

profit from the buy signals. We test various trading strategies using market and limit

orders and find that profits are available only when using limit orders to open and close

positions. The average abnormal return generated by using limit orders is 7.6 basis points

(bps) per day.

5

Using limit orders to buy and sell does not guarantee completed round

trips, and from 20% to 50% of our buy signals do not result in completed round trips. The

7.6 bps abnormal return uses only completed round trips.

Our basic results come from allowing our trading algorithm to process the

available data without questioning the sample of buy signals. In a more detailed analysis,

we clean the sample by removing all financial firms and by keeping only firms listed on

NASDAQ and the New York Stock Exchange (NYSE). NASDAQ firms are substantially

smaller than NYSE firms, and therefore valuable soft information should be more

concentrated at the headquarters of a NASDAQ-listed firm. We find larger positive

abnormal returns from NASDAQ-listed firms than from NYSE-listed firms, which is

consistent with a local information advantage.

Overall, our results are consistent with semistrong efficient markets and the

difficulty of transferring soft information. We find it difficult to extract local information

as we are only able to extract informative signals for buys. Using buy signals, it is

challenging to generate abnormal returns. Abnormal returns are available only when

limiting transaction costs by using limit orders to open and close the positions, which

exposes traders to the risk of not completing the needed trades. Additionally, we find

higher returns in NASDAQ-listed buy signals, which is consistent with a local

2

Quotes are visible to other traders on the open LOB market. On NASDAQ, full quoting data are available to

level 2 and 3 subscribers; level 1 subscribers may see only current best bid and best offer prices for a security.

3

We note that six months is a relatively short period. However, ITCH data include every message on

NASDAQ for a given period, so our six-month sample starts with about 41 billion observations.

4

These probabilities are significantly different from the 50% probability of a simple coin toss.

5

A 7.6 bps daily return annualizes into an abnormal return of about 19% over a 250-trading-day year. After

applying the well-known “day-of-the-week”effect, the daily and annual abnormal returns increase to 10.7 bps and

26.5%, respectively.

42 The Journal of Financial Research