What is it about?
Financial markets often react strongly to major government announcements, such as decisions made by the U.S. Federal Reserve (the Fed) about interest rates. These announcements, like those made during FOMC meetings, can cause stocks or bonds to rise or fall within minutes. But can we predict how the market will respond in advance? Our research explores this question. We developed a model that combines both the wording of key economic announcements and recent market data to forecast short-term market reactions. To improve accuracy, we trained the model using examples of real and hypothetical (counterfactual) events. This helps the model better understand what kinds of announcements actually drive market change. Our findings suggest that combining economic text with financial data can lead to more accurate forecasts and help investors better prepare for critical policy events.
Featured Image
Why is it important?
Our work is important for two main reasons. First, it helps investors make better trading decisions by forecasting how markets are likely to react to upcoming announcements. This is especially useful for event-driven investment strategies. Second, it provides a tool for policymakers to evaluate how their communications and decisions may influence financial markets. By simulating different versions of an announcement and analyzing the predicted impact, they can better understand the potential consequences before making them public. This makes our research both practically useful and timely in an era of rapid financial information flow and data-driven decision making.
Perspectives
Working on this paper has been an incredibly rewarding experience for me. As someone deeply interested in the intersection of artificial intelligence and economics, I’ve always been fascinated by how real-world events like FOMC meetings can trigger rapid market movements. This project gave me a chance to combine my passion for deep learning with the very human challenge of interpreting economic signals and their consequences. It was especially exciting to explore the use of counterfactual thinking, asking “what if” questions, to help models better understand the real impact of policy texts. I hope this work sparks further conversations among researchers, investors, and policymakers about how we can use AI to navigate the complexity of modern financial systems more transparently and responsibly.
Yang Zhang
Southwestern University of Finance and Economics
Read the Original
This page is a summary of: CAMEF: Causal-Augmented Multi-Modality Event-Driven Financial Forecasting by Integrating Time Series Patterns and Salient Macroeconomic Announcements, August 2025, ACM (Association for Computing Machinery),
DOI: 10.1145/3711896.3736872.
You can read the full text:
Contributors
The following have contributed to this page
