Trend forecasting in financial time series using acombinational method of heuristic patternrecognition and support vector machine

Whereas many studies have been done on forecasting various time series, it has always been associated with challenges such asuncertainty. For example, in financial time series, due to the time series' Non-Stationary feature, forecasting is likely toencounter false regression. To solve this problem, in this research, trend forecasting has been done instead of price forecasting.In this case, since the subtraction operator has been used to calculate the trend, the effect of the Non-Stationary feature isremoved and the issue of false regression is solved. To achieve this aim, the trend in financial time series has been predictedusing machine learning methods. In this research, the effective features of the last ۱۰ years in the commodity stock market datafor the shares of several various companies have been examined and compared with the market's benchmark index. Aftercreating different machine learning models and maximizing the accuracy of the results, a satisfying application has beenextracted to be used as an effective trading tool for traders. The Random Forests algorithm and Support Vector Machine,Feature Selection, and Heuristic algorithms have been used to train the model. The achieved results show that the proposedmodel is capable of producing accurate forecasts, and also outperforms other approaches currently in use.

Keywords:

Forecasting , Financial time series , Machine learning , Trend Prediction