TOWARDS A FUZZY VERSION OF AUTOMATIC TRADING SYSTEMS
TOWARDS A FUZZY VERSION OF AUTOMATIC TRADING SYSTEMS

Summary/Abstract: Since the stock markets exist, traders and stock-marketers dream of finding a formula or strategy that would allow them to detect in advance the best times to buy or sell assets. The detection of these signals would allow them to increase the profits by surpassing the average performance of the market, giving added value to their investment strategies. More recently, methods based on artificial intelligence, such as fuzzy systems, neural networks and genetic algorithms, grouped under the term "Soft Computing", have appeared in the field of finance. This paper proposes a fuzzy approach to designing automatic trading systems. Fuzzy logic is naturally adapted to integrate the vague aspects, omnipresent in the field of "trading". It is more flexible than a classical trading system, requires fewer rules, tolerates ambiguity and inaccuracy and finally, a fuzzy system is easily modifiable, simple to understand and also more modular than a tradition alone. Finally, our fuzzy trading system is applied to back-testing its performance on the historical data of the shares of the S.N.G.N. ROMGAZS.A. company (symbol: SNG), listed at Bucharest Stock Exchange.

• Details
• Contents