A0364
Title: Changing point prediction under multi-objective genetic algorithm
Authors: Mu-En Wu - National Taipei University of Technology (Taiwan)
ChiFang Chao - National Chengchi University (Taiwan) [presenting]
Ju-Fang Yen - National Taipei University (Taiwan)
Abstract: Technical analysis is a widely used investment approach, with the Elliott wave principle being one of its most prominent models. However, identifying Elliott waves in real-time remains challenging, as wave patterns often become clear only after market movements have already occurred. However, accurately identifying these reversal points is complex. Traditional methods rely on various technical indicators to generate buy and sell signals, yet they often lack a comprehensive quantitative framework. The integration of additional factors, such as overweighting mechanisms, stop-loss strategies, and machine learning techniques, is explored to enhance the precision of reversal point identification and improve overall returns. Using commonly applied technical indicators as feature labels, feature optimization is investigated for detecting trend reversals. The approach leverages a genetic algorithm-based feature selection framework combined with a strategy backtesting module to iteratively refine and evaluate profitability metrics. Experimental results from backtesting on the Taiwan stock market (2020-2022) indicate that the proposed method achieved a performance increase of 214.28\%.
