A hybrid deep learning model for accurate time series forecasting of cryptocurrencies

Abstract

A growing number of people and organizations are choosing to invest in cryptocurrencies. The development of precise forecasting models for cryptocurrencies is crucial due to their very volatile market. Financial forecasting has long made use of time series analysis and prediction, but conventional time series analysis techniques have trouble capturing intricate patterns and nonlinear relationships. On the other hand, although deep learning models show promise in time series analysis, their effectiveness depends on large amounts of data, which might result in overfitting. In order to predict bitcoin prices, this research presents a hybrid model that combines long short-term memory and convolutional neural networks. The CNN is used to extract features from the time series data, while the LSTM is used to capture persistent.