Trader types and fleeting orders: Evidence from Taiwan Futures Exchange

• Author: Ching‐Ting Lin,Wei‐Yu Kuo
• Date: 01 December 2018
• Published date: 01 December 2018
• DOI: http://doi.org/10.1002/fut.21963

Received: 14 June 2017

|

Accepted: 5 August 2018

DOI: 10.1002/fut.21963

RESEARCH ARTICLE

Trader types and fleeting orders: Evidence from Taiwan

Futures Exchange

Wei‐Yu Kuo

1

|

Ching‐Ting Lin

2

1

Department of International Business

and Risk and Insurance Research Center,

National Chengchi University, Taipei,

Taiwan

2

Department of Money and Banking,

National Chengchi University, Taipei,

Taiwan

Correspondence

Ching‐Ting Lin, Department of Money

and Banking, National Chengchi

University, 64, Sec. 2, Zhi‐Nan Road,

Wenshan District, Taipei 11605, Taiwan.

Email: ct.lin@nccu.edu.tw

Funding information

Ministry of Science and Technology,

Taiwan, Grant/Award Number:

105‐2410‐H‐004‐033‐MY3

Abstract

This paper investigates the relation between trader type and fleeting order

strategy based on unique data from Taiwan Futures Exchange. We find that

fleeting orders are more commonly used by institutional traders than individual

traders. There exists intraday seasonality of fleeting orders submitted by

institutional traders. Institutional traders do not chase market prices, but

respond to changes in immediate execution costs. Network orders constitute a

key component of the fleeting order strategy of proprietary traders. In sum,

different types of traders, including individual traders, proprietary traders, and

foreign traders, exhibit significantly different behaviors when deploying the

fleeting order strategy.

KEYWORDS

fleeting order, network trading, order cancellation, Taiwan futures markets, trader types

JEL CLASSIFICATION

G12, G14, G20

1

|

INTRODUCTION

On May 6, 2010, the US financial markets, including stock indices, stock index futures, options, and exchange‐

traded funds, fell rapidly by more than 5 %, followed by a quick and significant rebound within only 30 min. This

event is known as “flash crash.”Surveys indicate that algorithmic and high frequency trading are the main

contributors to flash crash events (Kirilenko, Kyle, Samadi, & Tuzun, 2016). Using algorithmic trading, high‐

frequency traders can engage in frequent, repeated, and ongoing interactions with the markets within

milliseconds, trade aggressively with price movements, and accelerate plunges in downward‐trending markets,

leading to a flash crash.

With improvements in trading technology, traders have increasingly used algorithmic trading strategies in recent

years (Hendershott, Jones, & Menkveld, 2010). Fleeting orders, defined by Hasbrouck and Saar (2009) as limit orders

cancelled within 2 s after submission, are one of the algorithmic trading strategies commonly used in the markets. They

document that as high as 93% of submitted limit orders in INET (an ECN acquired by Nasdaq in 2005) are cancelled or

revised. More interestingly, about 37 % of the cancellations are fleeting orders. At present, the practice of quick order

cancellation is widespread and has been of increasing concern to regulators. For example, CME Group exchanges

specifically prohibit manipulative order cancellation.

1

J Futures Markets. 2018;38:1443–1469. wileyonlinelibrary.com/journal/fut © 2018 Wiley Periodicals, Inc.

|

1443

1

The CME adopted new Rule 575 in September 15, 2014 to ban disruptive trading practices that were abusive to the fair execution of transactions. The new rule explicitly indicates that no person shall

place quotes with the intent to cancel the orders before execution to create a misleading appearance of market depth or price movements, known as “spoofing.”

Despite the increasing order cancellation intensity and regulatory concerns, empirical studies on fleeting orders

remain limited.

2

Most of them focus on the event of order revision and cancellation, but ignore the speed of its

occurrence. For example, Ellul, Holden, Jain, and Jennings (2007) show that 43% of orders submitted to the New York

Stock Exchange (NYSE) are cancelled. Liu (2009) and Fong and Liu (2010) find that the most aggressive orders account

for around 80% of cancellations across the entire limit order book of Australian stock markets. They also show that

order cancellation is highly correlated with monitoring costs. Specifically, the frequency of order cancellation decreases

when monitoring costs increase and vice versa. Raman and Yadav (2013) indicates that 24% of all incoming limit orders

in Indian stock markets are cancelled. He also shows that changes in market conditions and traders’inventories of held

stocks and correlated stocks influence cancellation behavior. Examining order cancellation in Taiwan Stock Exchange

(TSE), Chiao, Wang, and Tong (2017) document that foreign traders cancel limit orders most actively. However, they

are unable to unambiguously distinguish normal order cancellations from fleeting orders since trades in the TSE are

periodically executed every 15 s for each stock.

Hasbrouck and Saar (2009) directly examine fleeting orders. They establish three hypotheses for the motives behind

fleeting orders: Chasing, cost‐of‐immediacy, and search. In particular, the chasing hypothesis suggests that traders

cancel and place a more aggressive order when the same side order price moves away from the original quoted price.

The cost‐of‐immediacy hypothesis posits that traders cancel a limit order and switch to a market order for the reduced

cost of immediacy when the opposite side best price moves toward the original best price. Finally, the search hypothesis

postulates that searching for nondisplayed orders in the market is a motive of traders to submit fleeting orders.

Hasbrouck and Saar find evidence supportive of both the chasing hypothesis and the search hypothesis but against the

cost‐of‐immediacy hypothesis.

Since their data do not provide information about trader identity, Hasbrouck and Saar (2009) investigate the motives

of fleeting orders by evaluating the extent to which order cancellations are dependent on subsequent changes in the best

bid or ask prices without clearly identifying the sequence of orders submitted by the same trader. This lack of trader

identity, however, may result in ambiguity regarding the trading strategies used by traders. In particular, because they

cannot directly calculate the time interval between the submission and cancellation of a given order, Hasbrouck and

Saar indirectly use the life‐table method to define fleeting orders. Moreover, the sequences of order submission by a

given trader cannot be identified, so the dynamic trading strategy of a trader cannot be definitively recovered from the

data. Another issue is that since INET is primarily operated by market makers, the results in Hasbrouck and Saar (2009)

are more closely related to the trading behavior of market makers. Therefore, the fleeting order behavior of other types

of traders remains an unexplored topic.

The trader type could also be an important factor in the U‐shaped pattern of order cancellation documented in Liu

(2009) and Fong and Liu (2010), who show that monitoring cost is the main reason behind the pattern. If different types

of trader face different levels of monitoring costs, they may have different patterns of order cancellation and fleeting

orders. In addition, Raman and Yadav (2013) shows a positive relationship between order cancellations and network

trading, but does not provide any evidence on such a relationship for fleeting orders. Handa and Schwarz (1996) show

that traders can make profits by using a network trading strategy. By limiting risk with a predefined spread, a network

trading strategy could be used by traders to generate profits. Based on these findings, we therefore conjecture that

network trading strategy is probably correlated with fleeting order strategy. Such a relationship has not yet been well

documented in the literature. Hence, investigating the influences of trader type and network orders on fleeting order

strategy will be illuminating.

In this paper, we explore these issues based on a unique data set from Taiwan futures markets that contains all

submitted orders and completed trades of each trader.

3

Based on this data set, we are able to unambiguously identify

each trader and track his/her orders in this continuous auction market. We can also take snapshots of network trading

to study the relationship between fleeting orders and network trading strategy. Due to the variation of update frequency

of limit order book in the Taiwan Futures Exchange (hereafter TAIFEX), we define fleeting orders as the limit orders

cancelled within 5 s after submission.

4

To the best of our knowledge, this is the first paper that studies fleeting orders in

one of the major futures markets in Asia.

2

See Harris (1998), Bloomfield, O’Hara and Saar (2005), Large (2004), and Rosu (2009) for the theoretical foundations of order revision and cancellation.

3

On INET, traders are allowed to submit hidden limit orders that are not displayed for execution. However, traders in Taiwan futures markets are not allowed to submit hidden orders. We therefore do

not include the empirical results of the searching hypothesis in the paper but they are available from the authors upon request.

4

We also conduct the same empirical tests based on the fleeting orders defined by Hasbrouck and Saar (2009) as the limit orders cancelled within 2s after submission. The results are quantitatively

similar and reported later.

1444

|

KUO AND LIN