Enhancing Cryptocurrency Price Prediction through Inter-Coin Volatility and Hyperparameter Optimization
摘要
This paper presents a novel cryptocurrency price prediction method that boosts accuracy by combining Long Short-Term Memory (LSTM) networks with genetic algorithm-based hyperparameter optimization and integrating interdependencies among cryptocurrencies. Leveraging their volatility and complex dynamics, we employ volatility matrices and correlation coefficients to enhance predictions and market understanding. The genetic algorithm optimizes LSTM parameters, surpassing conventional methods. Assessed using multiple metrics, including Root Mean Squared Error (RMSE), our model proves robust across diverse cryptocurrency datasets and market conditions. These results underscore the value of merging deep learning with evolutionary algorithms, providing essential tools for market analysts and investors.