Developing an Investment Modeling Framework by Integrating Japanese Candlestick Patterns and Fuzzy Logic

Abstract

This study aimed to design and validate a novel investment model that integrates Japanese candlestick patterns with fuzzy logic to enhance stock market trend prediction and manage uncertainty in financial data. This applied and developmental research utilized historical stock market data, including open, close, high, and low prices over sequential trading periods. Data were transformed into candlestick representations, and features such as the Relative Strength Index (RSI), divergences, and failure swings were extracted. Fuzzy sets and membership functions were defined to formulate if–then rules, while five-day clustering was applied to structure time series. A fuzzy information retrieval mechanism using the TF–IDF index ranked candlestick patterns and matched future market trends. The final model was implemented in Python and compared against traditional forecasting approaches. Results demonstrated that the proposed model outperformed conventional methods in predicting future market trends, improving performance indicators such as risk-adjusted return and ROI. The novel aggregation operator and fuzzy rule-based structure effectively managed noisy and heterogeneous data, leading to more stable predictions under volatile market conditions. Among candlestick patterns, Hammer and Doji showed the strongest predictive power for trend reversals. Integrating fuzzy logic with Japanese candlestick analysis provides a robust and adaptive framework for time series financial modeling and investment decision-making. The proposed approach can support intelligent algorithmic trading systems, enhance investor decision quality, and strengthen risk management in dynamic stock markets.