Developing financial forecasting modeling with deep learning on silver / ounce parity

Metals such as gold, silver and platinum, which are accepted all over the world, experience fluctuations in their prices globally, and investors want to earn profits by evaluating these products. Technical and fundamental analysis methods are used for graphical direction estimation of these products. In addition, with the widespread use of deep learning methods, different methods have been put into practice for graphical direction estimation. Some of the most common of these methods are deep learning algorithms. In this study, LSTM architecture, which is one of the deep learning algorithms, and ARIMA architecture, which is a time series method, were used. With these architectures, financial forecasting models have been developed over silver/ounce parity. The training and test data were loaded into the established algorithms, allowing the system to learn, and silver/ounce parity prediction values for the next 10 days were produced. In order to increase the accuracy of the success of the algorithms, the algorithms were run 10 times and the average of the 10-day forecast data was taken. The reason for choosing this method is that it produces different predictions and graphs each time the algorithms are run. According to the results obtained in the research, ARIMA architecture produced better values than LSTM architecture.