Cryptocurrency price prediction and forecasting using machine learning algorithm and long- short term memory mapping

Abstract

Cryptocurrency has been a fascinating topic of many over the past few years, as many people are starting to trade these currencies for big cash-outs. It is a decentralized digital currency. It is a cryptocurrency that has used cryptography to manage the trading systems, creation, and management without relying on any third parties. Since the innovation of the first cryptocurrency Bitcoin in 2009, its value has skyrocketed. Starting from 0.09to42, 226 per bitcoin, it’s a very big market. New services, new companies are accepting it as it seems to be the new future. It is an encrypted peer-to-peer network for simplifying digital exchange. Blockchain-based currency has been the topic of many discussions and is widely popular lately. We decided to analyze the predictability of currency prices by using a machine learning algorithm widely known as the LSTM method. LSTM (Long Short-Term Memory) is a module provided for RNN, later developed and popularized by many researchers; like RNN, the LSTM also consists of modules with recurrent consistency and it works well with short time-sequential datasets. LSTM networks are well-suited for processing and making predictions. This machine-learning algorithm was developed to deal with vanishing gradient points encountered when training traditional RNNs and predict multiple currency prices for a short time interval. It will help traders and buyers to understand more about the price volatility of cryptocurrencies at one-minute intervals for real-life buy and sell.