2001年
Designing trading agents for an artificial market with a multi-objective genetic algorithm
ICCIMA 2001: FOURTH INTERNATIONAL CONFERENCE ON COMPUTATIONAL INTELLIGENCE AND MULTIMEDIA APPLICATIONS, PROCEEDINGS
- ,
- 開始ページ
- 226
- 終了ページ
- 230
- 記述言語
- 英語
- 掲載種別
- 研究論文(国際会議プロシーディングス)
- 出版者・発行元
- IEEE COMPUTER SOC
Artificial market(AM) is study of market economy through agent-based simulation. In AM, to construct agents having trading strategies with adequate rationality and variety is an intrinsic requirement. Difference of preference to return and risk among agents will be one candidate reason of variety of the trading strategies. In this paper, problem of constructing trading agent is considered as a two-objective optimization problem of minimizing risk and maximizing return, and is approached to by multi-objective genetic algorithm(MOGA). As for the artificial market environment, the U-Mart system proposed for a test bed of interdisciplinary study is used. Agents are evaluated in the U-Mart with other agents having simple and fixed strategies, and evolved with the MOGA. Computer simulation shows various agents having non-dominated trading strategies can be obtained by the proposed approach.
- リンク情報
- ID情報
-
- Web of Science ID : WOS:000173022300044