AI Powered Crypto Analytics Agent for Token Insights and Forecasting

Abstract

The cryptocurrency market is highly volatile and keeps changing rapidly, which makes accurate prediction and decision-making quite difficult. To handle this, we developed an AI-driven crypto analytics agent by combining time-series forecasting models like ARIMA, LSTM, and GRU with NLP-based sentiment analysis to gain real-time market insights.ARIMA worked well for capturing linear trends but struggled with sudden market fluctuations. On the other hand, deep learning models such as LSTM and GRU performed better by learning nonlinear patterns and long-term dependencies. We also used sentiment analysis to understand public mood from social media and news, adding context to numerical predictions.Data was collected from DexScreener, CoinGecko, and social media APIs, and the results were displayed on an interactive dashboard with features like whale transaction tracking, portfolio risk analysis, and sentiment-driven forecasting. Our evaluation showed that combining deep learning with sentiment analysis improved predictive accuracy and provided more practical insights for investors and analysts.

Keywords