Who’s Smart and Who’s Lucky? Inferring Trading Strategy, Learning and Adaptation in Financial Markets through Data Mining
Trading “profits” can be obtained by luck or by the implementation of a superior trading strategy. In this chapter we discuss the difficulties of distinguishing between the two. First, a suitable characterization of profit that distinguishes between trading gains and market gains is required. Secondly, one needs to be able to characterize trading “strategies”. To achieve this, we introduce the notion of a genotype-phenotype map to finance, where the genotype is associated with the information set and associated decision rules that lead to a given set of trading decisions for a given trader, while the phenotype is described by the set of observable trading decisions themselves. In AI based systems, such as agent-based markets, a strategy is implemented algorithmically and so the genotype is explicitly known. In real markets however, the genotypic trading strategy of one trader is hidden from the rest. The phenotype however, is, in principle, observable. A microscopic description at the level of the set of individual trades, however, is not sufficient to understand or characterize the strategies at a more macroscopic and intuitive one. By introducing a set of coarse grained variables that can be used to classify strategy types, we show how these variables can then be data mined to understand what differs between an intelligent and a lucky strategy. We show that these variables can be used to distinguish between different strategy types and can be further used to infer the presence of learning and adaptation in the market. We illustrate all of the above using data from an experimental political market.
KeywordsTrading Strategy Sharpe Ratio Limit Order Winning Strategy Real Market
Unable to display preview. Download preview PDF.
- 2.Farmer, J.D.: Market Force, Ecology and Evolution. Santa Fe Working Paper 98-12-117E Santa Fe Institute (1998)Google Scholar
- 4.Palmer, R.G., Arthur, W.B., Holland, J.H., Lebaron, B., Tayler, P.: Artificial Economic Life: A Simple Model of a Stock Market. Physica D 73 (1994)Google Scholar
- 5.Gordillo, J.L., Stephens, C.R.: Analysis of Financial Markets using Artificial Intelligence Techniques. Computación y Sistemas 6, 253–273 (2003)Google Scholar
- 6.LeBaron, B.: Agent-based Computational Finance. In: Tesfatsion, L., Judd, K.L. (eds.) Handbook of Computational Economics, vol. 2, ch. 24 (2006)Google Scholar
- 9.Bagehot, W.: [pseud], The Only Game in Town Financial Analyst’s Journal 27, 12–14 (1971)Google Scholar
- 10.Benink, H.A., Gordillo, J.L., Pardo, J.P., Stephens, C.R.: A Study of Neo-Austrian Economics Using an Artificial Stock Market, EFA 2004 Maastricht Meetings Paper No. 3218 (March 2004) (submitted to Journal of Empirical Finance), http://ssrn.com/abstract=567125
- 11.Stephens, C.R., Benink, H.A., Gordillo, J.L., Pardo-Guerra, J.P.: A New Measure of Market Inefficiency (August 24, 2007), http://ssrn.com/abstract=1009669
- 12.Hauser, F.: Die Presse Online-Wahlb~orse 2002 - Eine finanzwirtschaftliche Betrachtung. Thesis, University of Innsbruck (2003)Google Scholar
- 13.Hand, D., Mannila, H., Smyth, P.: Principles of Data Mining. MIT Press, Cambridge (2001)Google Scholar