Physics

AI System Predicts Financial Markets by Routing Text Through Global-Local Networks

How the science connects

Natural language p…Financial forecast…

AI Insight

GLoR-forecast is a novel forecasting method that combines global and local text routing mechanisms in a state-contingent manner for predicting financial and socio-economic time series. The approach dynamically selects and weights relevant textual information based on the current state of the time series, enabling more accurate predictions by incorporating context-specific signals from news, reports, and other text sources. The method demonstrates improved performance over traditional time-series models by leveraging the informational content of unstructured text data alongside numerical historical patterns.


This advancement could significantly improve economic forecasting, investment decision-making, and policy planning by better incorporating real-world events and sentiment captured in text. The state-contingent routing mechanism allows the model to adapt its information sources based on market conditions, potentially providing more reliable predictions during both stable and volatile periods.


Understand the Science

Natural language processing 77 articles Explore Concept → Financial forecasting Concept coming soon

Source: GLoR-forecast: state-contingent global-local text routing for financial and socio-economic time-series forecasting