INTIGRATED TIME-SERIES REAL-TIME CRYPTOCURRENCY PRICE PREDICTION USING HYBRID MODEL

The unpredictable and volatile nature of cryptocurrency markets has made accurate price forecasting an essential aspect of modern investment strategies. Traditional statistical models often fall short in capturing the complexity of these markets, primarily due to their assumption of linearity and the absence of adaptability to rapidly changing, non-stationary data. In response to these limitations, advanced computational techniques such as ensemble learning and deep learning have gained prominence for their ability to model complex patterns and relationships in time-series data. This research presents a novel comparative evaluation of deep learning architectures and ensemble methods for forecasting the prices of leading cryptocurrencies, including Bitcoin, Ethereum, Dogecoin, Binance Coin, Ripple, Solana, and Litecoin. Experimental findings reveal that models such as Gated Recurrent Units (GRUs), Simple Recurrent Neural Networks (RNNs), and Light Gradient Boosting Machines (LightGBM) consistently outperform both conventional machine learning algorithms and simplistic trading strategies like buy-and-hold or random walk. These insights not only highlight the strengths of modern predictive approaches but also provide investors with robust tools to enhance trading accuracy and reduce risk in the dynamic world of digital assets.