Competition Overview: A Continuation of a Top-Tier Global Financial Intelligence Event
Now in its fourth year, the AFAC Financial Intelligence Innovation Competition is open for global registration. The event is jointly initiated by nearly 30 top domestic and international universities, research institutions, and enterprises, including the China Computer Federation (CCF), Peking University, Fudan University, Nanyang Technological University (Singapore), Ant Group, and NVIDIA. According to official data, the competition has cumulatively attracted over 15,000 teams and nearly 50,000 participants from more than 600 universities and 400 companies, establishing itself as one of the most influential events in the financial AI domain.
This year’s competition aims to accelerate the practical application of AI and large model technologies in the financial industry by opening up real-world business scenarios and bridging the gap between academic research and industry needs.
Dual Tracks: For Both Algorithm Elites and Startup Teams
To cater to participants from diverse backgrounds, the competition features two parallel tracks: the “Challenge Track” and the “Startup Track”.
Challenge Track: Positioned as an algorithm-focused hackathon, this track is primarily for university students and developers worldwide. Participants will design and optimize algorithmic models to solve specific business problems using real, anonymized data from the financial sector provided by the organizers.
Startup Track: Introduced in 2024, this track focuses on discovering and nurturing early-stage projects with commercial potential, including One-Person Companies (OPCs) and startup teams. It encourages innovation in four main areas: “Smart Wealth Management and Banking Services,” “Smart Insurance and Risk Management,” “Inclusive Finance, Green Finance, and Pension Finance,” and “Frontier Technologies,” while also allowing participants to propose their own topics. Projects will be judged on creativity, technical innovation, and the feasibility, replicability, and scalability of their solutions.
Core Challenges: Four Frontier Tracks Targeting Key Industry Pain Points
The Challenge Track problems, jointly proposed by Fudan University and Tsinghua University, focus on four pressing technical challenges in the current FinTech landscape:
- Institutional Trading Behavior Recognition and Fund Flow Analysis: This challenge requires participants to analyze massive transaction datasets to identify the behavioral patterns of different institutional investors and track fund flows. It involves AI techniques such as time series analysis, Graph Neural Networks (GNNs), and anomaly detection, which are crucial for maintaining market stability and monitoring risks.
Complex Financial Document Reconstruction Challenge: The financial sector deals with a vast number of unstructured documents with complex formats (e.g., scanned prospectuses and annual reports). This track aims to use Document AI technologies, including Optical Character Recognition (OCR), Natural Language Processing (NLP), and layout analysis models, to accurately convert these documents into structured data for downstream model analysis.
Automated Experimentation Challenge with Sparse Feedback: In real-world financial products (such as credit approval or product recommendations), positive user feedback (e.g., purchases, conversions) is often very sparse. This challenge explores how to conduct efficient A/B testing and model iteration in data-sparse environments, potentially involving advanced methods like Multi-armed Bandits and Reinforcement Learning.
Financial Long-Text Agent Challenge: Dynamic Memory Compression and Efficient Q&A: This problem addresses the challenges faced by Large Language Models (LLMs) when processing extremely long financial documents (like regulatory filings and research reports). Participants need to design innovative dynamic memory and Retrieval-Augmented Generation (RAG) mechanisms to enable accurate and efficient Q&A on long-text content, building smarter financial analysis Agents.
Incentives and Schedule
The competition offers a total prize pool of one million RMB and provides a variety of incentives for winners in different tracks.
Challenge Track: In addition to cash prizes and certificates, winners will receive fast-track internship opportunities at major tech companies, Alibaba Cloud Tianchi computing power vouchers worth 300 RMB, and a chance to present their work at industry conferences.
Startup Track: The first-place prize is as high as 100,000 RMB. The competition also features a special “AFAC 2024 Top 20 Outstanding Startup Cases” award, providing winning teams with resources such as investor matchmaking, direct access to angel funds, and mentorship from venture capital experts.
Registration has recently opened globally. The initial project selection will take place in July 2024, followed by a project roadshow in mid-August. The final awards ceremony is scheduled for September in Shanghai at the Inclusion·The Bund Conference. Participants can register on the official competition website: https://afac.alipay.com.
