Abstract
We present an agent-based simulation of an asset market with heterogeneously informed agents. Genetic programming is applied to optimize the agents’ trading strategies. After optimization, insiders are the only agents able to generate small systematic above-average returns. For all other agents, genetic programming finds a rich variety of trading strategies that are predominantly based on exclusive subsets of their information. This limits their price impact and prevents them from making systematic losses. The resulting low noise renders market prices as largely informationally efficient.
Similar content being viewed by others
Notes
Meaning situations where no agent can enhance his payoff by changing his trading strategy.
With 10 signals ϵ i and two possible states for each signal, there are r = 210 = 1,024 possible states of the information system.
An exception are cases where CVo 50 = CVo 51. In those cases we follow Schredelseker (2001) by adjusting P to CVo 50 − 0.05 or CVo 50 + 0.05 if this increases the number of shares traded. I.e., when CVo 50 = CVo 51 = CVo 52, the median price would leave us with three neutral agents and a market volume of 48 shares. Instead, P = CVo 50 − 0.05 will allow for a market volume of 49 shares with 51 buyers facing 49 sellers. in this case, scale selling is used and buyers will only receive \(s= \frac{49}{51}\) shares.
As an agent in I = 0 gets no information at all, he will submit a reservation price of CV 0 = 5 when applying a fundamental strategy.
E.g., if an agent decides to buy the security, he will realize gains if the security is undervalued, and he will make losses if it is overvalued. If his decision does not influence the market price, he will have a 50% chance for gains and losses, leaving him with an average performance of 0.
The ramped half-and-half method proposed by Koza (1992) is used to ensure a rich diversity of trees.
As reservation prices smaller than zero or greater than ten make no sense in this model, fitness in such cases is set to the minimal possible value which corresponds to a return of −1.
This means that there is a 10% chance for reproduction, so that the selected strategy will be inherited by the new generation without any modification. This assures that successful strategies can survive.
See http://jgap.sourceforge.net/ for a documentation on this package.
The 100 agents are sorted in ascending order by their gains and one of the first agents is chosen using a log-normally distributed variable.
For a detailed discussion of this effect see Lawrenz (2008).
Note that every simulation of this kind is unique, so results on the development during optimization cannot be generalized. However, results after optimization do not change qualitatively when repeating the simulation or carrying out more runs. A robustness check where the simulation starts with different initial strategies supports this.
As indicated by the grey bars in Fig. 26, some of them have already been optimized, but the optimization did not derive a superior strategy for them.
Note that this is no equilibrium situation as described in Schredelseker (2001). In his model with three possible strategies, only 4 of 10 traders change strategies.
Note the different scaling of the chart when comparing it to Fig. 3.
The examples for trading strategies mentioned here are the simplest to be found after optimization. The formulas for the vast majority of optimized strategies are too complex to interpret.
Two strategies are considered to be identical if they derive identical returns in all 1,024 runs.
Note that in this optimization, the fundamental strategy of the insiders emerges endogenously. On average, genetic programming generates this strategy 520 times in each optimization step.
References
Brock WA, Hommes CH (1998) Heterogeneous beliefs and routes to chaos in a simple asset pricing model. J Econ Dyn Control 22:1235–1274
Fama EF (1970) Efficient capital markets: a review of theory and empirical work. J Finance 25(2):383–417
Figlewski S (1982) Information diversity and market behavior. J Finance 37:87–102
Glosten LR (1989) Insider trading, liquidity, and the role of the monopolist specialist. J Bus 62(2):211–235
Grossman SJ (1976) On the efficiency of competitive stock markets where traders have diverse information. J Finance 31:573–585
Hanke M, Schredelseker K (2009) Index funds should be expected to underperform the index. Appl Econ Lett Online First. doi:10.1080/17446540802599689
Hauser F (2008) Trader’s incentives to process information and market efficiency. In: Hanke M, Huber J (eds) Information, interaction and (in)efficiency in financial markets, Linde, pp 11–24
Hommes C (2006) Heterogeneous agent models in economics and finance. In: Tesfatsion L, Judd KL (eds) Handbook of computational economics, 1st edn, vol 2, pp 1109–1186
Huber J (2004) A contribution to solving an old puzzle: why different trading strategies persist in competitive markets. Journal of Academy of Business and Economics 3:171–190
Huber J, Kirchler M, Sutter M (2008) Is more information always better? Experimental financial markets with cumulative information. J Econ Behav Organ 65:86–104
Hule R, Lawrenz J (2008) The value of information. some clarifications and some new results for the schredelseker-game. In: Hanke M, Huber J (eds) Information, interaction and (in)efficiency in financial markets, Linde, pp 135–155
Jegadeesh N, Kim W (2006) Value of analyst recommendations: international evidence. J Financ Mark 9:274–309
Jeng LA, Metrick A, Zeckhauser R (2003) Estimating the returns to insider trading, a performance-evaluation perspective. Rev Econ Stat 85:453–471
Koza JR (1992) Genetic programming: on the programming of computers by means of natural selection. The MIT Press, Cambridge, Massachusetts
Lawrenz J (2008) Understanding non-monotonic payoffs for heterogenously informed agents. In: Hanke M, Huber J (eds) Information, interaction and (in)efficiency in financial markets, Linde, pp 115–134
Lux T (1995) Herd behavior, bubbles and crashes. Econ J 105:881–896
Lux T, Marchesi M (2000) Volatility clustering in financial markets, a microsimulation of interacting agents. Int J Theor Appl Financ 3:675–702
Madura J, Ngo T (2008) Impact of ETF inception on the valuation and trading of component stocks. Appl Financ Econ 18:995–1007
Pfeifer C, Schredelseker K, Seeber GU (2009) On the negative value of information in informationally inefficient markets: calculations for large number of traders. Eur J Oper Res 195:117–126
Schredelseker K (1997) Zur Ökonomischen Theorie der Publizität. In: Ott C, Schäfer HB (eds) Effiziente Verhaltenssteuerung und Kooperation im Zivilrecht, J.C.B. Mohr, pp 214–245
Schredelseker K (2001) Is the usefulness approach useful? Some reflections on the utility of public information. In: McLeay S, Riccaboni A (eds) Contemporary issues in accounting regulation. Kluwer Academic Publishers, pp 135–153
Spiegel M, Subrahmanyam A (1992) Informed speculation and hedging in a noncompetitive securities market. Rev Financ Stud 5(2):307–329
Acknowledgements
We thank Michael Hanke, Jürgen Huber, Klaus Schredelseker, all discussants at the workshop “Evolution and Market Behavior in Economics and Finance”, and three anonymous referees for very helpful comments on this paper. We also acknowledge the FNR (Fonds National de la Recherche Luxembourg) for financial support on this project.
Author information
Authors and Affiliations
Corresponding author
Electronic supplementary material
Appendix
Appendix
Rights and permissions
About this article
Cite this article
Hauser, F., Kaempff, B. Evolution of trading strategies in a market with heterogeneously informed agents. J Evol Econ 23, 575–607 (2013). https://doi.org/10.1007/s00191-011-0232-6
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00191-011-0232-6