Hanwha Life announced on November 19 that it presented a jointly developed “AI-based arbitrage model” with Stanford University’s Human-Centered AI (HAI) at ICAIF 2025, the world’s most prestigious academic conference in financial AI, held in Singapore from November 15 to 18.
ICAIF (International Conference on AI in Finance) is organized by the Association for Computing Machinery (ACM) and is regarded as one of the leading global gatherings for finance-focused AI research. Major institutions such as JPMorgan, Morgan Stanley, and BlackRock, along with academic researchers from around the world, participate annually.
This year, 349 papers were submitted, and 113 (32.4%) were accepted after peer review. Hanwha Life’s submission ranked in the top 15.5 percent and was selected for the Oral Presentation session, highlighting its recognition as a high-quality research contribution.
The paper, titled “Attention Factors for Statistical Arbitrage,” applies the attention mechanism—widely used in modern generative AI—to traditional financial factor models. While attention identifies meaningful signals in large datasets, factor models are used to detect common drivers behind stock price movements.
The proposed model demonstrated strong performance using historical U.S. equity market data, achieving a high Sharpe Ratio and proving its effectiveness in generating returns relative to investment risk.
The research also utilized deep learning to predict residual time series—price deviations between stocks that should theoretically move together. This prediction capability allowed for more precise portfolio adjustments, improving real-world returns even after accounting for transaction costs.
A Hanwha Life spokesperson stated, “This study shows that AI can detect subtle signals overlooked by traditional financial models, revealing new investment opportunities. We will continue expanding application-driven research that connects technology directly to investment performance.”
The ICAIF review committee noted that the study “lays a key foundation for future financial AI research.”
The joint project was conducted by the Hanwha Life AI Research Lab and Professor Markus Pelger of Stanford University’s financial engineering department. The code and sample data will be released on GitHub, and the full paper will be available on arXiv. Hanwha Life plans to broaden global collaboration and contribute to the financial AI research ecosystem by actively sharing its research.
