Predatory Algorithms in High Frequency Trading

24 settembre 2014
Versione stampabile

Time: 16:30
Location: A222 - (Floor Below DISI Povo1), via Sommarive - Povo-Trento


  • Prof. Julian Williams, Durham University, Business School

High frequency trading (HFT) and the degree of automation of financial markets has stimulated a very active debate on how financial assets and contracts are traded. Several studies of equity transactions on the US national market system (NMS) have indicated that since the advent of Regulation-NMS upwards of 70% of all transactions are instigated and cleared by fully autonomous algorithms. We can classify these algorithms into a variety of phyla; however the broad terms passive versus predatory are commonly used. Passive algorithms act as clearing mechanisms "arbitraging" opportunities from misalignment of quotes in the limit order-book at high speed and maintaining a zero average position. Some algorithms are designed to extract patterns from "technical-analysis" to provide positive returns to investment funds often across asset classes (as classic example is high-speed spot-future arbitrage). These strategies assume that the market dynamics are exogenous to their activity 
and passively wait and act as opportunities arise. However, another less "friendly" set of algorithms, the "predators", exist that are designed to drive markets in particular directions by taking advantage of imperfections in liquidity, transactions costs and asymmetric information. We can think of these algorithms as "malware-for-markets". They take advantage of the gap between the theory of how markets work, the resultant implementation of that theory into electronic matching engines and the actual observed results of the deployment of the technology of the market in-vivo. This objective of this talk is to provide the interested non-specialist in the background and provide some importance sampling on new and interesting topics in this area and hopefully provide some ideas for research on detecting predatory HFT algorithms.

About the Speaker:
Julian Williams is professor of accounting and finance at Durham University Business School since January 2014. He has published widely on the topics of financial regulation, risk management and portfolio management. His main interests lie in market microstructure, regulating complex derivative securities and the impact of the liquidity of these instruments on the cost of capital for individuals, companies and governments. He has also cooperate with computer scientist on economic aspects of security (and is scientific director of the SECONOMICS EU project). Julian's work has been commented on and referenced in the Financial Times, the Press and Journal and OECD publications.

Professor Julian Willians will be visiting UNITrento 'till the end of September.

Contact: fabio.massacci [at] (Fabio Massacci)