An end-to-end Python implementation of Cao et al.'s (2025) HLPPL methodology for the identification of financial (asset price) bubbles. Implements 7-parameter Log-Periodic Power Law model fitting, confidence-weighted sentiment analysis, regime-dependent 'BubbleScore' fusion, and Transformer-based forecasting with a backtesting framework.
deep-learning sentiment-analysis pytorch transformer quantitative-finance algorithmic-trading bert asset-pricing risk-management backtesting time-series-forecasting econophysics financial-modeling finbert market-prediction nlp-finance behavioral-finance research-replication financial-bubbles lppl-model
-
Updated
Oct 16, 2025 - Jupyter Notebook