Sharks in the Dark: Quantifying Latency Arbitrage

54 Pages Posted: 9 Mar 2017 Last revised: 27 Sep 2021

See all articles by Matteo Aquilina

Matteo Aquilina

Bank for International Settlements

Sean Foley

Macquarie University

Peter O'Neill

UNSW Australia Business School, School of Banking and Finance

Thomas Ruf

University of New South Wales (UNSW)

Date Written: September 5, 2016

Abstract

Using proprietary order book data with participant-level message trac
and matching engine time stamps, we investigate stale reference pricing in
dark pools. We document a substantial amount of stale trading which im-
poses large adverse selection on passive dark pool participants. We show that
HFTs almost never provide dark liquidity, instead frequently consuming dark
liquidity, in particular in order to take advantage of stale reference prices. Fi-
nally, we examine several market design interventions to mitigate stale trades,
showing that only mechanisms to protect passive dark liquidity, such as ran-
dom uncrossings, are e ective at ensuring accurate reference prices.

Keywords: high-frequency trading, dark pools, latency arbitrage, reference prices

JEL Classification: G10, G14, G18

Suggested Citation

Aquilina, Matteo and Foley, Sean and O'Neill, Peter and Ruf, Thomas, Sharks in the Dark: Quantifying Latency Arbitrage (September 5, 2016). FCA Occasional Paper No. 21, Available at SSRN: https://ssrn.com/abstract=2848120

Matteo Aquilina (Contact Author)

Bank for International Settlements ( email )

Basel
Switzerland

Sean Foley

Macquarie University ( email )

North Ryde
Sydney, New South Wales 2109
Australia
0417702600 (Phone)

Peter O'Neill

UNSW Australia Business School, School of Banking and Finance ( email )

Sydney, NSW 2052
Australia

HOME PAGE: http://peteroneill.org

Thomas Ruf

University of New South Wales (UNSW) ( email )

Kensington
High St
Sydney, NSW 2052
Australia

Do you have negative results from your research you’d like to share?

Paper statistics

Downloads
372
Abstract Views
2,024
Rank
138,511
PlumX Metrics