PASCAL - Pattern Analysis, Statistical Modelling and Computational Learning

A Novel Strategy for the 2005 Penn–Lehman Automated Trading Competition
Rahul Savani and Ben Veal
(2005) Technical Report. Centre for Discrete and Applicable Mathematics, LSE, London, UK.

Abstract

The Penn-Lehman Automated Trading (PLAT) project is a broad investigation of algorithms and strategies for automated trading in financial markets. It is centered on the Penn Exchange Simulator, which simulates an electronic crossing network. The simulator realistically mixes bids from automated stock trading agents with bids from the real stock market. The participants in the project develop automated trading agents that are entered in formal automated trading competitions. This paper concerns the 2005 PLAT competition. Unlike previous years, this competition was run with automated agents only, and no real-time stock market data. This allowed the authors’ winning strategy, Jump & Dump, to exploit a basic, but easily overlooked, flaw in the other agents. We outline Jump & Dump and explain how and why it won.

EPrint Type:Monograph (Technical Report)
Project Keyword:Project Keyword UNSPECIFIED
Subjects:Theory & Algorithms
ID Code:1441
Deposited By:Martin Anthony
Deposited On:28 November 2005