The Research on the Prediction of Cryptocurrency Based on Linear Regression and LSTM

Abstract

As the growing interest of investment on Cryptocurrencies and the huge volatility of their price, a need for scientific model to predict the future price is growing.In this context, the paper uses linear regression and LSTM model to predict the price of Bitcoin and Ethereum.The result shows that the prediction made by Linear Regression shows less errors but greater lag compared with the prediction made by LSTM method.The lag problem is considered to generate from lack of peripheral information other than previous prices.The prediction implies that the prices of Cryptocurrencies are theoretically predictable, and shows a direction of further research, such as the use of mixed-LSTM model.The methods provided in this paper can be used in development of better models and further investments.