Knowledge discovery in financial investment for forecasting and trading strategy through wavelet-based SOM networks

Abstract

The stock market has been a popular financial investment channel in the recent era of low interest rates. How to maximize profits is always the main concern for investors; and different investors have different preferences about the holding periods of their investments. In this study, in contrast to other related studies, we propose a hybrid approach on the basis of the knowledge discovery methodology by integrating K-chart technical analysis for feature representation of stock price movements, discrete wavelet transform for feature extraction to overcome the multi-resolution obstacle, and a novel two-level self-organizing map network for the underlying forecasting model. In particular, a visual trajectory analysis is conducted to reveal the relationship of movements between primary bull and bear markets and help determine appropriate trading strategies for short-term investors and trend followers. The forecasting accuracy and trading profitability of the proposed decision model is validated by performing experiments using the Taiwan Weighted Stock Index (TAIEX) from 1991 to 2002 as the target dataset. The resultant intelligent investment model can help investors, fund managers and investment decision-makers of national stabilization funds make profitable decisions.