A NEW TIME-SERIES CHART PATTERN RECOGNITION APPROACH

摘要

Time-series chart pattern recognition has attracted arnlot of research interests, especially in stock technicalrnanalysis from historical patterns. The paper brings a na?vernefficient algorithm to do backward screening forrndetermining acceptable chart patterns by applying thernwell-known pattern “Cup with Handle” to optimizernIntelligent Stock Selector System (ISSS), a rule-basedrnintelligent system. Bezier Curve and statistics were usedrnto test the results. The contribution of this research is tornfacilitate data mining based modeling system design forrnextensive forecasting and research purposes for financialrninstitutions such as investment banks, proprietary tradingrnfirms and hedge funds, as well as individuals.