Forecasting Intraday USD/CAD Exchange Rate with News-Derived Monetary-Policy Signals

2026-09-26

AI Summary

This study addresses the limitation of existing exchange rate forecasting research, which focuses predominantly on the aggregate sentiment of monetary policy communication while neglecting attribution analysis of specific transmission channels. To bridge this gap, this work proposes an interpretable signal decomposition and attribution framework that leverages large language models to transform news into structured signals. By integrating temporal feature engineering with tree-based models, the framework evaluates the incremental predictive value of each dimension through rolling-window experiments under false discovery rate control. The findings demonstrate that multidimensional signals significantly outperform single sentiment indicators in forecasting efficacy. Furthermore, communication timing is identified as the strongest predictor, validating the effectiveness of non-sentiment signals and underscoring the central role of attribution analysis in quantifying news-derived signals for exchange rate prediction.

Abstract

Monetary-policy announcements and central-bank communications play a central role in foreign exchange markets, yet their qualitative, unstructured form makes their forecasting value difficult to quantify. While prior research has largely focused on sentiment extracted from financial news, comparatively little is known about the relative contribution of different dimensions of monetary-policy communication. Existing studies primarily evaluate whether textual information improves overall forecasting performance but provide limited insight into which communication channels drive such improvements. To address this gap, this paper introduces a statistical attribution methodology that decomposes monetary-policy communication into interpretable channels and quantifies their incremental forecasting contribution under false-discovery-rate control. Monetary-policy news is transformed into structured communication signals using large language models (LLMs) and temporal feature engineering. These signals are evaluated using rolling-window experiments with tree-based machine-learning models. The results show that monetary-policy communication contains measurable predictive information. Attribution analysis shows that predictive value is concentrated in a small subset of signals, with communication timing providing the strongest individual feature-level contribution, targeted communication-activity measures also contributing positively, and LLM-derived sentiment providing complementary information at the group level. The findings indicate that communication-based forecasting value extends beyond sentiment alone and that attribution, rather than aggregate accuracy alone, is central to evaluating news-derived signals.

Problem

• monetary-policy communication
• exchange rate forecasting
• signal attribution
• predictive contribution
• intraday USD/CAD

Innovation

• Large Language Models
• Statistical Attribution
• False Discovery Rate Control
• Monetary-Policy Signals
• Tree-based Machine Learning