Adaptive sniping for volatile and stable continuous double auction markets

Toft, I. E. and Bagnall, A. J. (2009) Adaptive sniping for volatile and stable continuous double auction markets. Agent-Mediated Electronic Commerce and Trading Agent Design and analysis, 13. pp. 119-134.

Full text not available from this repository. (Request a copy)

Abstract

This paper introduces a new adaptive sniping agent for the Continuous Double Auction. We begin by analysing the performance of the well known Kaplan sniper in two extremes of market conditions. We generate volatile and stable market conditions using the well known Zero Intelligence-Constrained agent and a new zero-intelligence agent Small Increment (SI). ZI-C agents submit random but profitable bids/offers and cause high volatility in prices and individual trader performance. Our new zero-intelligence agent, SI, makes small random adjustments to the outstanding bid/offer and hence is more cautious than ZI-C. We present results for SI in self-play and then analyse Kaplan in volatile and stable markets. We demonstrate that the non-adaptive Kaplan sniper can be configured to suit either market conditions, but no single configuration is performs well across both market types. We believe that in a dynamic auction environment where current or future market conditions cannot be predicted a viable sniping strategy should adapt its behaviour to suit prevailing market conditions. To this end, we propose the Adaptive Sniper (AS) agent for the CDA. AS traders classify sniping opportunities using a statistical model of market activity and adjust their classification thresholds using a Widrow-Hoff adapted search. Our AS agent requires little configuration, and outperforms the original Kaplan sniper in volatile and stable markets, and in a mixed trader type scenario that includes adaptive strategies from the literature.

Item Type:	Article
Faculty \ School:	Faculty of Science > School of Computing Sciences
UEA Research Groups:	Faculty of Science > Research Groups > Data Science and Statistics
Depositing User:	Users 2731 not found.
Date Deposited:	25 Sep 2012 12:45
Last Modified:	10 Jan 2024 01:21
URI:	https://ueaeprints.uea.ac.uk/id/eprint/39672
DOI:	10.1007/978-3-540-88713-3_9

Actions (login required)

View Item