遺傳模糊專家系統（GA-Fuzzy）技術在證券投資操作之決策輔助

近年來模糊專家系統在證券市場投資之決策輔助上已日漸受重視。模糊專家系統乃透過與人腦思維「過程模糊、結論清晰」的特徵，以近似人類思維方式的推論過程來獲致精確的結果。一般而言這些方法較為符合投資人在投資過程中常運用的「若… 則… 」之定性處理模式。本研究主要在於提出一套以遺傳演算法為基底的模糊專家系統，以遺傳演算法來動態學習技術指標中隱含的資訊，並將此資訊轉化為符合投資者使用之投資策略，進而作為投資者買進或賣出之交易參考。實際資料乃取自台灣股價加權歷史指數作為驗證模式的基礎。結果顯示GA-Fuzzy 專家系統建議之最適投資策略可以增進投資績效。

The applications of fuzzy expert systems in aiding stock market investment are increasingly emphasized in recent years. Fuzzy expert systems mimic human reasoning approach that synthesizes approximate input information into crisp and precise decision. This advantage helps mass investors using the If-Then-Else terms of rules to support the stock trading decision. This research proposes a Genetic Algorithm-based approach to construct the fuzzy expert system by inferring the historical trading data implicitly. Several candidate technical indexes are considered in the fuzzy rules set to guide trading strategies. By adopting the Genetic Algorithms approach, the selection of appropriate indexes along with related fuzzy rules and corresponding membership functions can be effectively determined. The learning data derived from Taiwan’s stock market is used for developing the fuzzy rules. A prototype GA-Fuzzy system is developed and tested. The results indicate that the system is able to improve the investment performance.