Forecasting KOSPI based on a neural network with weighted fuzzy membership functions

This paper presents a methodology to forecast the direction of change in the daily Korea composite stock price index (KOSPI) using input features that are derived from KOSPI and KRW/USD exchange rates by technical indicators and the non-overlap area distribution measurement method based on the neural network with weighted fuzzy membership functions (NEWFM). To extract input features used in NEWFM, three numbers of technical indicators, including RSI (Relative Strength Index), CCI (Commodity Channel Index), and CPP (Current Price Position), are selected in the first step. In the second step, thirteen numbers of input features and an input feature derived by three numbers of technical indicators (RSI, CCI, and CPP) and a technical indicator (CPP) are extracted from the daily KOSPI and KRW/USD exchange rates, respectively. In the final step, the minimized numbers of input features are selected by the non-overlap area distribution measurement method from fourteen numbers of input features to forecast the next day’s direction of change in the daily KOSPI. The best performance result of NEWFM is 59.21% using thirteen numbers of input features selected by the non-overlap area distribution measurement method.