Financial monetary economy
milad asadpour; ALIREZA Rezaee
Abstract
The economy and the forecasting of its indicators is one of the main and influential elements in the life of every person and can be the basis for the superiority of individuals and governments over others, as a result, the forecasting of indicators is always one of the main challenges and concerns of ...
Read More
The economy and the forecasting of its indicators is one of the main and influential elements in the life of every person and can be the basis for the superiority of individuals and governments over others, as a result, the forecasting of indicators is always one of the main challenges and concerns of economists and investors. On the other hand, with the emergence of cryptocurrencies and the increase in their price value, a lot of investment has been made in this field, so finding methods to predict the price of cryptocurrencies in the future is very important. In this article, a method for predicting the price of Bitcoin using artificial intelligence algorithms is presented. For this purpose, the characteristics affecting the future price of bitcoin were identified and categorized and standardized in two separate data sets including price data and structural data of the bitcoin network, then a new structure consisting of three feedforward and feedback neural networks was designed, the first and second network including the GRU layer. which predict prices in parallel and separately from each other. Next, the output of each of these networks is combined with each other by a neural network, and finally, values are obtained as price predictions for the coming days. The results of the research and error calculation show that using the last 15 or 20 days is the best interval for predicting the future price of Bitcoin, which has an accuracy of 97.36% and 96.76%, which shows the high accuracy and efficiency of this method. Also, due to the correct selection of the input features, the computational volume of this research has been significantly reduced.
