The chart represents result of testing for a new upgraded NNSTP-2*. To test the software, an artificial input data have been created. The data is a sum of three function: a linear one with k = 0.05 and two sinusoidal functions – T1 = 360 days (amplitude 20), T2 = 40 days (amplitude 5).
The tool exhibits an average deviation between actual curve (green curve) and predicted one (pink curve) 0.07%. Yellow curve presents an extrapolation beyond known data.
*) Neural networks can discover patterns in data that humans might not notice and successfully predict the future trend. Neural Network Stock Trend Predictor NNSTP-2 predicts future share prices (or percentage changes) using Fuzzy Neural Network. Input data are daily weighted closing price of the stock and the volume traded. Forecast horizon can be chosen by user from one day to the limit that defined by downloaded historical data period: (predicted days) = [(historical days) - 40] / 20