Forecasting of digital currency prices using deep learning algorithm

Abstract

Cryptocurrencies are still a novelty in the financial market which is not fully explored yet. This makes cryptocurrencies an interesting and relevant topic for researchers, regulators and investor. The cryptocurrency market is constantly growing, volatile and ever-changing, making it extremely risky for investments. Financial literacy, long-term prescience and correct focus without getting lost in chaos of risk, value and unpredictability are especially important in today's economy. Due to complex nature of the cryptocurrency market the large amount of profile data is generated, which can be processed using a variety of methods such as statistical, technical, sentiment, solution and other analysis. Previous studies have used two main method for cryptocurrency prices prediction: empirical and machine learning algorithm analysis. The aim of this article is to test the applicability of LSTM deep learning algorithm for cryptocurrency price forecasting as well as to evaluate the results. The research results showed that deep learning algorithm are able to forecast with low error magnitude, moreover trend of predicted values is accurate when compared to observed values. The results suggest that deep learning algorithm application for forecasting of cryptocurrency market price is a suitable alternative for risk minimization.