A109 金融商品の価格学習エージェント群への情報提示の方法と市場のマクロ構造の変化の関連に関する研究(マルチエージェント金融市場)

概要

論文の詳細を見る
In the field of economics, it is said that a market price shows random-walk behavior and cannot be expected. The theory of "Efficient Market Hypothesis" is the basis of this behavior. This theory is considered that all traders are homogeneous. However, it is appropriate to a trader is heterogeneous. And price fluctuation like a bubble phenomenon in recent years is the phenomenon which cannot explain sufficiently in the economics. Then, the artificial market research by the agent base simulation is appearing as new approach for solving such a phenomenon. There are many factors, such as economy, politics and a social problem, and a relation of demand supply, in change of a price. Then, in order to understand a market, both microscopic and macroscopic analysis is required. Based on such a situation, in our study, we build the artificial market model with the factor abstracted as a dealing signal. And we analyzed the influence which the difference in the composition by the agents has on a market. In an artificial market, agent who plays the role of trader expects a price using the model similar to a classifier. The condition part of the classifire is 10 bits, and consists of 2 bits represent the information that it is common between agents, and 8 bits represent the state of a market. An agent receives the prediction value of the next rate of price fluctuation from the classifier. And the optimal amount of financial-assets possession, an order price, and the amount of orders are calculated, and an order is taken out to a market there. An agent calculates the optimal amount of financial-assets possession, an order price, and the amount of orders from the value. The simulation was performed using two kinds of agents. One is a common information agent and another is an individual information agent. And we measured the complexity of a market using the technique of correlation dimension analysis. Consequently, the artificial market which consists of only agents who trade with common information does not have a certain structure and has the feature near a random walke. The artificial market which consists of only agents who trade with individual information has a certain structure and can be described in determinism. The artificial market constituted by the agents trade with common information and the agents trade with individual cannot be completely described in determinism, although it has a certain structure. Since the actual market also had the same feature, it is suggested that trader does not refer to only common information and only individual information in actual market.
社団法人日本機械学会の論文
2001-11-14

A109 金融商品の価格学習エージェント群への情報提示の方法と市場のマクロ構造の変化の関連に関する研究(マルチエージェント金融市場)

スポンサーリンク

概要

著者

関連論文

スポンサーリンク