This is the first study to systematically connect fine-grained argumentative structure with forecasting success, highlighting the role of structured reasoning in predictive accuracy.
📄 Read the full paper at: tinyurl.com/argstrategies

🧠 Forecasts that show balanced argumentative coverage, addressing multiple possible hypotheses rather than focusing on a single one, are associated with higher accuracy.
Graph-based models also outperform text-only baselines.

🔍 We model forecasters’ reasoning as argument graphs capturing support, conflict, and balance between propositions.
These graphs are enriched with text embeddings and analysed using a Graph Neural Network (GNN) to link structure and outcome.

📈 Accurate forecasting depends not only on data quality but also on the structure of reasoning behind predictions.
We examine how the organisation of arguments (how forecasters present, balance, and connect claims) relates to forecasting accuracy.

Debela Gemechu, Ramon Ruiz-Dolz, John Lawrence and Chris Reed: Practical Solutions to Practical Problems in Developing Argument Mining Systems
aclanthology.org/2025.argmini...

Debela Gemechu, Ramon Ruiz-Dolz, Kamila Górska, Somaye Moslemnejad, Eimear Maguire, Dimitra Zografistou, Yohan Jo, John Lawrence, Chris Reed: The Open Argument Mining Framework
aclanthology.org/2025.acl-dem...

Ramon Ruiz-Dolz, Zlata Kikteva, John Lawrence: Mining Complex Patterns of Argumentative Reasoning in Natural Language Dialogue
aclanthology.org/2025.acl-lon...

Debela Gemechu, Ramon Ruiz-Dolz, Henrike Beyer and Chris Reed: Natural Language Reasoning in Large Language Models: Analysis and Evaluation
aclanthology.org/2025.finding...