Forecasting Crypto Market Prices Using Stacked Bidirectional LSTM

The crypto market refers to the marketplace for cryptocurrencies, which are digital or virtual currencies that rely on cryptography to ensure security and prevent counterfeiting. This market has significantly influenced global financial systems, introducing decentralised finance and blockchain‑based transactions. By offering faster, more transparent, and borderless financial operations, it has revolutionised the traditional financial industry and challenged conventional banking and payment methods. Amid the rising geopolitical and economic challenges, global currency values have declined, stock markets have struggled, and investors have faced losses. This has renewed the interest in digital currencies. Due to the decentralised nature of cryptocurrency networks, predicting their prices is challenging, given their complexity, lack of central authority, and high market volatility. Our objective is to accurately forecast cryptocurrency price fluctuations to support profitable investments. This study employs Long Short‑Term Memory (LSTM) networks, a deep learning approach, to predict prices, focusing on Ethereum and Bitcoin using reliable historical data. Experimental results indicate that the projected model outstrips other algorithms in terms of mean absolute error (MAE), mean square error (MSE), and overall accuracy. This model aims to help investors reduce the financial risks and make informed decisions.