Financial Forecasting

Stories that (are) Move(d by) Markets: A Causal Exploration of Market Shocks and Semantic Shifts across Different Partisan Groups
We demonstrate that semantic shifts in news language can be causally linked to financial market shocks, with partisan differences influencing predictive power. Text signals prove particularly valuable during unexpected events like COVID-19.
Felix Drinkall, Janet B. Pierrehumbert, Stefan Zohren
PDF Cite DOI
When Dimensionality Hurts: The Role of LLM Embedding Compression for Noisy Regression Tasks
Demonstrates that compressing LLM embeddings can improve performance on noisy regression tasks like financial prediction by reducing overfitting, suggesting compression acts as a regularization mechanism.
Felix Drinkall, Janet B. Pierrehumbert, Stefan Zohren
PDF Cite DOI