Fuzzy Logic based System for Identifying Trading Relationships

Abstract

Applying Fuzzy Logic to Identify Trading Patterns The major concern of this study is to develop a system that can identify relationships between entities participate in the stock markets by analyzing historical trading data. Stock markets are complex. During each and every second, thousands of transactions happen on each finance exchange which makes millions of turnover to that country. Their dramatic movements and unexpected booms and crashes, make all traditional rule-based market surveilling tools useless and as result of that they does not provide the required capability of identifying relationships between users of the system during offline surveillance on historical data. Due to the vulnerabilities of the procedures and the lack of tools to identify them, financial frauds or malpractices may take place. This study attempts to resolve such complexity on detecting abnormal trading behavior and their relationships using the subtractive clustering based fuzzy system identification method. This study uses Mamdani Fuzzy Inference System Modeling to classify relationships and the system implementation is based on client-server architecture where both are implemented using C++. The model presented in this study elicits, from historical trading data, some of the rules which govern the market, and shows that rules which are drawn from a particular stock are to some extent independent of that stock, and can be generalized and applied to other stocks regardless of specific time or industrial field to identify or recognize trading patterns. Experimental test results of this study reveals that the model can correctly identify the trading patterns among different entities of the market with a firing threshold indicating the strength of participation in entity instances on identified relationship on the transaction.