BaseRock Research & Developing
develops proprietary AI for real-time market forecasting. We fuse minute-level price data with sentiment from social media and news to deliver adaptive, actionable predictions — validated on Bitcoin today, with a modular architecture built to extend to other crypto assets and financial markets.
Forecasting Systems
About NextTrend
Bitcoin pricing exhibits extreme non-stationarity, heteroskedastic volatility, and decoupling from traditional valuation. NextTrend is a high-frequency forecasting architecture that combines autoregressive price signals with natural language features. Using cloud-native infrastructure for minute-level inferencing, it deploys a hierarchical stacked ensemble. A Temporal Fusion Transformer (TFT) integrates a NeuralProphet-LSTM time-series branch with a regime-switching sentiment branch (FinBERT and ModernBERT). The approach achieves a MAPE of approximately 0.31% over a 12-hour horizon and significantly outperforms a random walk baseline, demonstrating the efficacy of attention-based mechanisms in capturing the interplay between social signals and price discovery. While Bitcoin is the primary validation case today, the same modular design is built to extend to other crypto assets and financial markets.
Read Article
Forecasting Bitcoin with Hybrid Sentiment Price Ensembles: Evidence from NextTrend Project
Forecasting Bitcoin’s price remains one of the most complex challenges in modern finance. Without traditional valuation anchors and with prices that shift on the wave of global sentiment, conventional models struggle to keep pace. The NextTrend project confronts this challenge by building a real-time forecasting system that fuses quantitative market dynamics with natural language insights from social media and news. The system blends NeuralProphet for long-term trend capture, LSTM for short-term volatility correction, and an XGBoost model that translates sentiment and engagement patterns into predictive signals. Working together, these components generate twelve-hour forecasts that react to both technical momentum and the tone of the online conversation. Deployed on Google Cloud, the pipeline ingests minute-level data, retrains continuously, and visualizes results through an integrated dashboard. The combined ensemble significantly improves predictive accuracy over standard baselines and random walk benchmarks, offering traders a transparent and adaptive lens on the world’s most sentiment-driven asset.
Have any questions? Get in Touch