Prediksi Cryptocurrency Berbasis LSTM Menggunakan Multi Modal Indikator Trading(Studi: Ethereum dan Solana)

Abstract

The dynamic development of the cryptocurrency market causes digital asset prices to experience high volatility, making it difficult for investors to accurately predict price movements. Therefore, an analytical method is needed to model price movement patterns in time series data. This study aims to develop a cryptocurrency price prediction model for Ethereum and Solana using the Long Short-Term Memory (LSTM) method with a multi-modal trading indicator approach. The dataset used consists of historical price data including open, high, low, close, trading volume, and technical indicators such as Exponential Moving Average (EMA), Relative Strength Index (RSI), and Bollinger Bands. The research process follows the CRISP-DM methodology, which includes business understanding, data understanding, data preparation, modelling, evaluation, and deployment stages. The data were processed through normalization and time series windowing, with a training and testing data split of 80:20. The evaluation results using Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE) indicate that the model has good predictive performance. The Ethereum model produced an RMSE value of 129.08 and a MAPE of 3.26%, while the Solana model produced an RMSE of 8.30 and a MAPE of 3.63%. The developed model was also implemented in a Streamlit-based dashboard to visualize prediction results interactively, helping users monitor and analyze cryptocurrency price movements.