【CFP】Call for Participation in Competitions @ ICAIF 2025 in Singapore 15 - 18 Nov
Fengbin Zhu
C1: FinSurvival Challenge: Advancing Deep Survival Modeling for Financial Transactions
Organizers: Oshani Seneviratne (RPI), Fernando Spadea (RPI), Kristin Bennett (RPI), Adrien Pavao (Codabench, MLChallenges), Aaron Green (RPI), Corey Curran (RPI)
Description: Survival modeling, the prediction of when a particular event will occur, has long been a cornerstone of statistical analysis in healthcare and engineering. In finance, it is equally powerful, with applications such as estimating time to loan repayment or default, forecasting portfolio churn, and anticipating customer attrition. These capabilities provide critical insights for risk management, strategic planning, and retention. The FinSurvival Challenge brings survival modeling to the world of DeFi transaction data. Participants are tasked with building models for time-to-event prediction: given an index event (e.g., a loan issuance), predict the time until a specific outcome (e.g., repayment or liquidation). Performance will be evaluated primarily using the Concordance Index (C-index), where 0.5 indicates random guessing and 1.0 represents perfect foresight. The goal is to advance beyond the existing FinSurvival benchmark, leveraging modern machine learning and deep learning approaches to capture the rich, non-linear patterns present in real-world financial data.
Website: https://finsurvival.github.io
C2: ACM ICAIF 2025 Cryptocurrency Forecasting Competition
Organizers: Yihao Ang (National University of Singapore), Qiang Wang (National University of Singapore), Yifan Bao (National University of Singapore), Xinyu Xi (National University of Singapore), Anthony K. H. Tung (National University of Singapore), Qiang Huang (Harbin Institute of Technology (Shenzhen)), Hao Ni (University College London), Lukasz Szpruch (University of Edinburgh)
Description: Cryptocurrency markets operate without the circuit-breaker protections or overnight cooling-off periods that are typical in equity markets. This 24/7 trading environment, combined with comparatively shallow order books and a retail-dominant investor base, leads to significant short-term volatility and frequent regime shifts, making short-horizon forecasting both difficult and essential. Cryptocurrency forecasting differs substantially from (1) generic time series prediction, where series are often stationary and exogenous signals are optional, and from (2) traditional financial forecasting, where valuations are anchored in firm fundamentals such as earnings and cash flows. Participants are challenged to predict next-hour log-returns for the top traded cryptocurrencies using historical returns and volumes. This competition bridges predictive modeling and market microstructure, enabling robust models that support algorithmic trading, risk management, and broader market transparency and inclusion.
C3: Secure FinAI Contest 2025
Organizers: Shengyuan Lin (Columbia University, Carnegie Mellon University), Keyi Wang (Columbia University), Jiechao Gao (Columbia University, Stanford University), Yupeng Cao (Stevens Institute of Technology), Yan Wang (The Fin AI), Kairong Xiao (Columbia University), Xiao-Yang Liu Yanglet (Columbia University)
Description: Financial AI (FinAI) is being increasingly applied in financial tasks such as trading, financial statement analysis, and compliance. The Secure FinAI Contest 2025 aims to advance the development of secure and reliable FinAI agents. It consists of three tasks: 1) FinRL-transformer for cryptocurrency trading, 2) FinGPT agents for SEC filings analysis, and 3) FinGPT agents for regulation and compliance. Participants will be provided with second-level limit order book data for cryptocurrency trading and high-quality question sets for benchmarking LLMs. The contest not only challenges participants to develop secure and innovative FinAI solutions but also advances FinAI by focusing on high-stakes regulation and compliance fields.
Website: https://github.com/Open-Finance-Lab/SecureFinAI_Contest_2025
C4: FinDDR 2025: Financial Document Deep Research Challenge
Organizers: Fengbin Zhu (National University of Singapore), Chao Wang (6Estates Pte Ltd), Chang Liu (Asian Institute of Digital Finance), Shuo Zhang (Bloomberg), Ke-Wei Huang (Asian Institute of Digital Finance), Huanbo Luan (6Estates Pte Ltd), Tat-Seng Chua (National University of Singapore)
Description: Financial analysis is crucial for informed decision-making among stakeholders of public companies. Yet extracting insight from lengthy and complex annual reports remains a significant challenge. Utilizing the proven capabilities of Deep Research Agents, we propose the Financial Document Deep Research (FinDDR) Challenge, which adopts similar deep research methodologies in question design and evaluation frameworks. The FinDDR Challenge introduces a richly structured, industry-diverse dataset and requires participants to generate comprehensive, sectioned research reports. This is accomplished through a hierarchical, stepwise reasoning framework that closely emulates the analytical methodologies employed by professional financial analysts. In conclusion, the FinDDR Challenge seeks to establish new benchmarks for complex document-based deep research in financial AI applications, fostering progress and collaboration across both academic and industry communities.
Website: https://finddr2025.github.io/
C5: Agentic Retrieval for Financial Documents: AI Grand Challenge on FinAgentBench
Organizers: Yoon Kim (MIT), Alejandro Lopez-Lira (University of Florida), Dhagash Mehta (BlackRock), Edward Tong (Google), Jacob Chanyeol Choi (LinqAlpha), Zach Golkhou (J.P. Morgan), Yongjae Lee (UNIST), Atlas Wang (UT Austin & XTX Markets), Igor Halperin (Fidelity Investments), Andrew Chin (AllianceBernstein), Sy Bor Wang (National University of Singapore), Joo Won Lee (Arrowpoint Investment Partners), Chee Seng Chan (Universiti Malaya), Lixin Fan (WeBank), Georgios Papaioannou (Qube Research & Technologies), Jihoon Kwon (LinqAlpha), Minjae Kim (LinqAlpha), Juneha Hwang (LinqAlpha), Suyeol Yun (LinqAlpha), Jin Kim (LinqAlpha)
Description: The competition challenges participants to build agentic retrieval systems that answer complex institutional finance questions grounded in raw SEC filings with character-level evidence attribution. Built on FinAgentBench, the task requires two-stage reasoning: (1) selecting the most relevant document type (10-K, 10-Q, 8-K, DEF-14A, or earnings call transcript) and (2) identifying supporting passages within that document. The benchmark includes 2,400+ expert-curated QA-evidence pairs from 2023–2024 filings. Hosted on Databricks Free Edition with optional Upstage Solar LLM integration, the challenge emphasizes interpretability, auditability, and reproducibility. Co-hosted with the AI for Finance Symposium ’25, it brings together academia, industry, and regulators to advance trustworthy financial AI systems.
Website: ai4f.org
Competitions Chairs:
Fuli Feng (University of Science and Technology of China)
Kunpeng Zhang (University of Maryland)