How Aggressive Are High‐Frequency Traders?
Author | Björn Hagströmer,Lars Nordén,Dong Zhang |
Published date | 01 May 2014 |
DOI | http://doi.org/10.1111/fire.12041 |
Date | 01 May 2014 |
The Financial Review 49 (2014) 395–419
How Aggressive Are High-Frequency
Traders?
Bj¨
orn Hagstr¨
omer
Stockholm University School of Business
Lars Nord´
en∗
Stockholm University School of Business
Dong Zhang
Stockholm University School of Business
Abstract
We study order aggressiveness of market-making high-frequency traders (MM-HFTs),
opportunistic HFTs (Opp-HFTs), and non-HFTs. We find that MM-HFTs follow their own
group’s previous order submissions more than they follow other traders’ orders. Opp-HFTs
and non-HFTs tend to split market orders into small portions submitted in sequence. HFTs
submit more (less) aggressive orders when the same-side (opposite-side) depth is large, and
supply liquidity when the bid–ask spread is wide. Thus, HFTs adhere strongly to the tradeoff
between waiting cost and the cost of immediate execution. Non-HFTs care less about this
tradeoff, but react somewhat stronger than HFTs to volatility.
Keywords: high-frequency trading, aggressiveness, order submission, liquidity, volatility
JEL Classifications: G14, G18
∗Corresponding author: Stockholm University,School of Business, S-106 91 Stockholm, Sweden; Phone:
+46 8 6747139; Fax: +46 8 6747440; E-mail: ln@fek.su.se.
We would like to thank NASDAQ OMX Group Inc. for providing the data, and Petter Dahlstr¨
om and
Frank Hatheway for numerous discussions about the data and market structure. Helpful comments from an
anonymous referee, Jonathan Brogaard, Michael Goldstein (the guest editor), and seminar participants at
Lund University (Arne Ryde Workshopin Financial Economics), Gothenburg University, and University
of London (Seventh International Conference on Computational and Financial Econometrics), are also
gratefully acknowledged. All three authors are grateful to the Jan Wallanderand Tom Hedelius foundation
and the Tore Browaldhfoundation for research support.
C2014 The Eastern Finance Association 395
396 B. Hagstr¨
omer et al./The Financial Review 49 (2014) 395–419
1. Introduction
In modern equity markets, traders differ in technology,sophistication, and strat-
egy. Retail and institutional investors coexist with proprietary traders who use com-
puters to pursue strategies with subsecond investment horizons. These traders may
base their decisions on the limit order book, which conveys the current state of liq-
uidity supply and demand to its participants. Limit orders may be characterized by
their aggressiveness, which the trader chooses by setting the limit price that leads
to immediate or to some extent delayed expected execution. Collectively, the ag-
gressiveness of all limit orders forms the supply and demand of liquidity. Order
aggressiveness is thus fundamental to the market quality variables that are central to
market microstructure research.
In this paper, we examine the interaction between limit order aggressivenessand
trader type. Specifically, we categorize limit orders and the limit order book by the
level of aggressiveness and investigate empirically how order aggressiveness differs
between traditional traders and different types of high-frequency traders (HFTs). By
conditioning order aggressiveness on the state of the order book as well as the actions
of other traders, we provide new insights into the functioning of electronic limit order
book markets.
The theoretical literature shows that order aggressiveness is determined by the
tradeoff between waiting costs and the cost of immediate execution.In Parlour (1998),
traders choose to submit limit orders or market orders based on their observation of
the limit order book depth. Foucault (1999) suggests that market orders are more
costly when the asset volatility is higher. Foucault, Kadan and Kandel (2005) model
how traders’ impatience determines their order placement strategies, and Ros¸u (2008)
holds competition in liquidity supply as a key variable for order aggressiveness. The
cost of immediacy is the same for all traders, but waiting costs may differ with the
sophistication of the trader. For example, an HFT may have the means to evaluate
patterns in order submissions of other traders before deciding what orders to submit,
whereas a slow trader does not have the capacity for such an analysis. Empirical
evidence of such differences are provided by Hendershott and Riordan (2013), who
show that algorithmic traders have an edge in revising their orders quickly following
public news announcements.
Several empirical articles analyze how order submission strategies depend on
the state of the order book (Griffiths, Smith, Turnbull and White, 2000; Handa,
Schwartz and Tiwari, 2003; Ranaldo, 2004) and the actions of other traders (Biais,
Hillion and Spatt, 1995). Due to data restrictions, most empirical analyses focus on
trading and quoting strategies at an aggregate level.However, as order submissions are
increasingly automatic, it is important to understand the differences between human
and different types of automatic order submissions. Hendershott and Riordan (2013)
study the interaction of order submissions between human traders and algorithmic
traders at the Deutsche B¨
orse. Brogaard (2010) analyzes the segment of algorithmic
traders that perform proprietary trading only, referred to as HFTs, and focuses on the
To continue reading
Request your trial