Real-Time Cryptocurrency Price Prediction using ARIMA and LSTM Models

Abhishek Pandey |
abhishekpandey0041gmail.com
Student, Computer Science and Engineering, Parul University, Vadodara

Abstract:Cryptocurrency markets are highly dynamic and volatile due to their decentralized nature and sensitivity to market sentiment, regulatory changes, and technological developments. Accurate prediction of cryptocurrency prices is therefore a challenging yet important task for investors and financial analysts. This research focuses on real-time cryptocurrency price prediction using two widely used time series forecasting techniques: the statistical AutoRegressive Integrated Moving Average (ARIMA) model and the deep learning based Long Short-Term Memory (LSTM) model. In this study, real-time cryptocurrency data is collected using financial APIs such as CoinGecko, consisting of historical price attributes including open, high, low, close prices, and trading volume. Data preprocessing techniques such as missing value handling, normalization, and stationarity testing are applied to improve model performance. Exploratory Data Analysis (EDA) is performed to understand trends and volatility patterns in cryptocurrency price movements. The ARIMA model is implemented as a baseline forecasting technique for linear time series data, while the LSTM model is used to capture nonlinear patterns and long-term dependencies. The performance of both models is evaluatedusing Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Squared Error (RMSE). Experimental results indicate that the LSTM model provides better prediction accuracy compared to ARIMA due to its ability to model complex patterns in highly volatile data. This research demonstrates that deep learning approaches such as LSTM provide more effective solutions for real-time cryptocurrency forecasting compared to traditional statistical methods.
Keywords: Cryptocurrency, ARIMA, LSTM, Time Series Forecasting, Deep Learning, Real-Time Prediction.