Daily Digital Currency Values Estimation Using Artificial Intelligence Techniques

Abstract

Recently, with the rapid rise in crypto money prices, Bitcoin has begun to be seen as an investment tool. Because of this trend, predictions in the digital money market gain importance. For this reason, in this study, a machine learning model was developed that can make daily predictions for Bitcoin, the most important currency in the digital currency market. An artificial neural network was used to make daily predictions for Bitcoin and the data set was designed with values from the coinmarketcap site. The next day's close price is estimated by using the open, high, low, volume, marketcap features from this site. In this study, unlike other studies, the closing price of the next day was tried to be estimated. Thus, a model has been developed that makes a value estimation that the investor will need. While creating the data sets, 300 days of data were used. In addition, considering the changes in the Bitcoin market, 3 different data sets were created as easy, moderate and hard. In this study, 0.9949, 0.9908 and 0.9503 R values were obtained in the test data sets of easy, moderate and hard difficulty levels, respectively. 70% of the data set was used for training. 15% of the data set was used to test the model. The remaining samples were used for validation. Considering the results obtained in the study, it was concluded that the estimation of Bitcoin closing values can be made daily using machine learning methods. In addition, it has been observed that there is a serious decrease in success rates on days when the price changes are too much.