The third edition builds on the 2024 (Brooklyn) and 2025 (Singapore) ICAIF workshops and focuses on the next generation of agentic, tool-augmented, retrieval-grounded, and governance-aware generative AI systems for high-stakes financial workflows.
About the Workshop
The workshop moves beyond isolated model capabilities toward reliable institutional systems, covering agent orchestration, harness engineering, workflow-level evaluation, red teaming, retrieval and tool use, governance, auditability, model-risk management, and human-agent collaboration. Applications span investment research, compliance, credit, risk, ESG, reporting, surveillance, and quantitative strategies, supported by new datasets, benchmarks, and infrastructure.
As LLMs move from single-turn text generation toward agents that plan, retrieve, reason, use tools, and coordinate with humans, financial institutions face a new class of system-level challenges around reliability, auditability, compliance, and operational risk. The 2026 edition sharpens the focus on governed agentic AI systems for institutional finance: reliable, auditable, and deployable systems for real institutional workflows.
Speakers
2026 THEME
Authority, Not Capability: Who Vouches for AI in Finance?
AI can now write the code, draft the research note and work through the mathematics. This year’s workshop asks what evidence, and whose judgment, makes those outputs safe to rely on for real financial decisions. Every session takes the same question: where do you still insist on a human, and why exactly there?
Keynotes
Huyên Pham, PhDProfessor of Applied Mathematics; President, Bachelier Finance Society
Rama Cont, PhDProfessor of Mathematics; Head, Oxford Mathematical and Computational Finance Group
Invited Talks
Raffaella Calabrese, PhDChair of Statistics and Data Science
Mathieu Chalvidal, PhDQuantitative Researcher
Panel 1
Fundamental
The Analyst and the Agent: LLM Agents That Read Filings, Earnings Calls and Broker Research
Analysts and portfolio managers on where agents now read disclosures and commentary for them, and where they still check every line.
Thierry CiszewskiHead of Equity Capability
Jérôme LegrasManaging Partner, Head of Research
Kurt Ze’ev KaraHead of Global Value Equities & AI
Florian SutterPortfolio Manager
Beatrice CicerchiaEquity Strategist
Marton Peter PriceCo-founder and Fund Manager, EverestQuant
Founder, Everesteer
- Moderator
Jacob Chanyeol Choi, PhDCo-founder & Co-CEO, LinqAlpha
Panel 2
Quant
Signals and Sign-Offs: When Agents Propose, Code and Backtest Trading Signals
Quant heads on agents that generate signal ideas, write code and run backtests, how to tell a real signal from an overfitted one, and who approves it for live capital.
Robert De WittManaging Director, Head of Quantitative Strategies Group
Markus Ebner, PhDHead of Multi Asset
Gianpaolo Tomasi, PhDDirector, QIS Research
Maurizio Morini, PhDSenior Financial Data Scientist
Adjunct Professor, University of Bologna
Mishel QyranaQuant
Ayub Hanif, PhDExecutive Director, Head of European Quantitative Strategy
- Moderator
Edward TongStaff Data Scientist, Google
Panel 3
Build
Built to Be Checked: Data Lineage, Permissions and Logs Behind Auditable AI Agents
Data and AI leaders from banks, asset managers and data providers on what must sit underneath an agent before a risk committee or supervisor accepts it.
Nicolas MochHead of SEBx
Alexander HorvatAI Solutions, Advisory & Financing Solutions
Shuyi YangHead of GenAI & ML Engineering
Nico BaumIndependent practitioner and researcher, AI and financial innovation
Bhaskarjit SarmahHead of AI Research for Financial Services
- ModeratorJacob Chanyeol Choi, PhDCo-founder & Co-CEO, LinqAlpha
Saturday 14 November 2026, Bocconi University
Registration and coffee
Opening remarks
Jacob Choi, LinqAlpha
Keynote
Generative AI for Financial Dynamics: From Scenario Generation to Risk Decisions
Huyên Pham, École Polytechnique
Panel 1
Fundamental: The Analyst and the Agent
Coffee and Poster Session I
Invited talk
LLMs for litigation forecasting
Raffaella Calabrese, University of Edinburgh
Panel 2
Quant: Signals and Sign-Offs
Oral papers I
Five peer-reviewed talks
- 12:10Could This BE Any More Wrong? Teaching VLMs to Know What They Don’t Know in Document Extraction (Varad Srivastava)
- 12:20Multidimensional Views Behind Target-Price Revisions: An LLM-Based Representation of Analyst Outlooks (Kiyoun Kim)
- 12:30Can You Trust Your Finance Agent? Drift, Diagnosis and Auditable Intervention on One Shipped Financial Research Agent (Cosimo Galeone)
- 12:40Using LLMs for Table Data Extraction from Public Finance Documents (Vikram Aggarwal, Aditi Mascarenhas)
- 12:50When Filings Are Silent: LLM News Readings and Volatility Forecasts Around Mergers (Donghwa Seo)
Lunch and Poster Session I (continued)
Invited talk
Risk in motion: large covariance estimation in the age of generative AI
Mathieu Chalvidal, Capital Fund Management
Panel 3
Build: Built to Be Checked
Oral papers II
Two peer-reviewed talks
- 15:00Deep Research Agents for Investment Surveillance in Information-Sparse Markets (Dhagash Mehta)
- 15:10Buy the Rumor, Sell the News: When Is News Priced In? (Arman Khaledian)
Keynote
Financial Tomography: Design and Validation Principles for Generative models in finance
Rama Cont, University of Oxford
Coffee and Poster Session II
Oral papers III
Four peer-reviewed talks
- 16:30ZooBench: Localizing Failure in AI Replication of the Factor Zoo (Sebastian Rivera)
- 16:40ClimInvestBench: Evaluating Agentic Investment Report Generation and Maintenance (Ed Richards)
- 16:50Right Numbers, Wrong Questions: Evaluating LLMs on Underspecified Financial Queries (Radha Yadav)
- 17:00Correct but Unlicensed: Prompt Allowlists and the Traceability of Figures in LLM Financial Assistants (Hyeonwoo Jeon)
Closing remarks
Networking and Poster Session II
until 18:00
Programme as of 10 November 2026, subject to change.
Oral Presentations
Can You Trust Your Finance Agent? Drift, Diagnosis and Auditable Intervention on One Shipped Financial Research Agent
Cosimo Galeone, Giuseppe Ettorre, Anna Ettorre, Ginevra Mariani, Minsu Park, Daniele Ligorio
Buy the Rumor, Sell the News: When Is News Priced In?
Alireza Kargarzadeh, Nariman Khaledian, Navid Parvini, Sid Ghatak, Arman Khaledian
Correct but Unlicensed: Prompt Allowlists and the Traceability of Figures in LLM Financial Assistants
Siddhesh Chaubal, Debarya Dutta, Hyeonwoo Jeon
ZooBench: Localizing Failure in AI Replication of the Factor Zoo
Campbell Taylor, Sebastian Rivera, Sami Naeem, Alejandro Canete
Deep Research Agents for Investment Surveillance in Information-Sparse Markets
Tianjiao Zhao, Jingrao Lyu, Kevin Zhao, Aslan Taheri, Marlin Brown, Tao Chen, Yueqi Wu, Dhagash Mehta
Right Numbers, Wrong Questions: Evaluating LLMs on Underspecified Financial Queries
Radha Yadav, Prasanna Devadiga, Peeyush Mishra, Apoorv Singh
ClimInvestBench: Evaluating Agentic Investment Report Generation and Maintenance
Edward Samuel Richards, Javier Sanz-Cruzado Puig, Richard McCreadie
Could This BE Any More Wrong? Teaching VLMs to Know What They Don’t Know in Document Extraction
Varad Srivastava, Prabhu R, Victor Daniel Casas Hernandez
Multidimensional Views Behind Target-Price Revisions: An LLM-Based Representation of Analyst Outlooks
Kiyoun Kim, Sung-Yeon Nathan Lim
Using LLMs for Table Data Extraction from Public Finance Documents
Vikram Aggarwal, Siddarth Raman, Susan Thomas, Jay Kulkarni, Aakriti Narang, Aditi Mascarenhas, Ajay Shah
Posters
A Discussion on Generative Preference Elicitation in Finance
Anagha Tiwari, Brian Jabarian, Alexander G. Gray, Nick Feamster, Alex Imas, Alex Kale
A Holistic Evaluation of Large Language Models for Earnings Forecasting
Yurun Sun, Pietro Panzarasa, Valentin Danchev
AI For Everyone: Measuring Assumptions and Mapping a Curriculum in Financial Services
Nidhi Singh, Will Flanagan, Rajitha Ramanayake, Sarah Epstein, Ivan Wiryadi, Gisele DeWitt, Ananya Nagireddy, Ana Alfaro Pizana, James Usher
Attribution Is a Propagation Problem in Multi-Agent Trading Systems
Advije Rizvani, Giovanni Apruzzese, Pavel Laskov, Nils Lukas
BTFM: Learning and Transferring Financial Representations from Public Transaction Histories
Rodrigo Mendoza-Smith, Yassine Landa, Will High
Can Personalized Advice Improve Investor Behavior? Evidence from a Multi-Agent Simulation of Retail Investors
Yu Jeong Lee, Sujin Jung, Jaeho Jo, Junyoung Kwon, Sangjin Jin, Yongjae Lee
CLAIM: A Framework for Auditing LLM-Generated Financial Measures
Xuan Yao, Kelvin J.L. Koa, Salsabila Fatma Aripa, Dai Yang, Yuxuan Tan, Zhou Yi, Ke-Wei Huang
Cross-Lingual Reusability of LLM Pipelines: Financial Extraction, Multilingual Reasoning, and Semantic Screening
Joong Ho Choi, Sherine George, Jiayang Zhao, Jayesh Rathi, Ashwin Kandath, Ananya Sane, Vishhvak Srinivasan
DealsFlow: An Evidence-Gated Agentic Harness for Venture Capital Deal Triage
Ming Ze Tang, David Cutting, Boyd Carson, Vasiliki Carson
Decision-Focused Representation Adaptation for Cold-Start IPO Future-Peer Retrieval
Yeonsuk Jung, Jang Ho Kim
Disagreement About Nothing: A Pre-Registered Stress Test of a Persona-Ensemble Measure of 10-K Complexity
Aryan Chhabra, Kuntara Pukthuanthong
Emergent Trading Biases in LLM Agent Swarms: Behavioral-Finance Benchmarks and Information-Channel Ablation
Kian Jagtiani
Evaluating LLM-Directed Model Development: Lessons from Income Prediction
Fang Fang, Alexander Statnikov
Extracting ESG Fields from Emerging-Market Sustainability Reports with LLM Assistants: A Pilot Evaluation on Kazakh and Uzbek Corporate Disclosures
Feruza Kholmukhamedova
FinALTA: Deterministic Execution with Language-Model Explanations for Auditable Financial Question Answering over Budget Data
Areej AlOtaibi, Layali Abdulrahman Alsolami, Tasniem Nasser Alyahya
FinCausalAudit: Separating Error Detection from Evidence Use in Financial LLMs
Maximilian Nicholson
FinPanel: Selecting LLM Trader Personas for Financial Trading
Boming Yang, Yuquan Wang, Yusuke Iwasawa, Yutaka Matsuo, Jiaxian Guo
FINSABER-2: A Multimodal Benchmark for Comparing LLM Trading Agents with Passive and Quantitative Strategies
Feiyang Huang, Weixian Waylon Li, Yaxuan Kong, Mihai Cucuringu, Tiejun Ma
From Letter-of-Credit Requirements to Executable Specifications: Evaluating Verification Accuracy and Completeness
Ji Sub Choi, Whan Choi, Seo Yeon Gu, Kim Suhyun, Inyong Yun
Grounded Event Extraction from SEC 8-K Filings with a Fine-Grained Taxonomy
Rian Dolphin, Joe Dursun, Jarrett Blankenship, Katie Adams, Quinton Pike
Honest Small Decision Models for Finance: Scale, Specialization Collapse, and Mixed-Corpus Remediation
Ajitabh Kumar
Is All the Information in the Price? LLM Embeddings versus GICS and Price-Correlation Stock Clusters in a Factor-Model Test of the Efficient Markets Hypothesis
Tucker Balch, Maria Hybinette, Grant Johnson, Bingyang Wang
LLM-Generated Counterfactual Stress Scenarios for Portfolio Risk Simulation via Hybrid Prompt-RAG Pipeline
Masoud Soleimani
MemTrial: Learning When to Trust Memory in LLM Portfolio Agents
Guanghao Wu, Zhuo Cai, Shoujin Wang
Multi-Agent LLMs for Replicating, Auditing, and Extending Corporate Bond Trading Strategies
Zachary Alfalahi
Reading the Market: Portfolio Crash Prediction via Narrative Tracking and News-Coverage Bursts
Ziling Liu, Kelvin J.L. Koa, Ke-Wei Huang
Semantic Lifting for Institutional Digital Asset Taxonomies: An Auditable LLM Workflow for DeFi Classification
Hyunho Jeon, Seongchan Park, Byungsoo Kang, Gaeun Kim, Daehyun Kim
Ten Thousand Rows Is Not Enough: An Evaluation Protocol for Foundation-Model Return Signals
Kuan Ta Wu
The Expert and the Machine: A Case Study of Human-Steered and Autonomous Agents Iterating a Niche Domain Simulator
Ollie Olby, Andreea Bacalum, Kibaek Lee, Namid Stillman
The Price of Thought: Does Test-Time Reasoning Pay in LLM Trading?
Jiayi Chen, Guiling Wang
Training Scores Are Not Returns: Workflow-Level Evaluation of LLM-Guided Alpha Search
Siyan Chen
Universal Diffusion Models for Implied Volatility Surfaces: Learning Shared Dynamics Across Stocks
Mingzhi Yang, Sheng Wang, Chao Zhang, Ruikun Li
Verify Before You Sign: Benchmark Sensitivity and Tool-Output Deference in LLM Portfolio Advice
Nithilan Karthik, Arjun Subramanian
When Filings Are Silent: LLM News Readings and Volatility Forecasts Around Mergers
Donghwa Seo, Juyeong Lee, Doohwi Cha, Minsuk Sung, Minjae Lee, Seunghan Son, Yoontae Hwang
When Repeatable Labels Disagree: Prompt Reformulation in LLM Measurement of Financial Disclosures
Hongzhen Du
Your AI, On a Dial: Controlling Investment Bias in LLMs with a Single Neuron
Sahong Park, Suhwan Park, Hoyoung Lee, Gakyung Kwon, Wonbin Ahn, Jaewon Choi, Alejandro Lopez-Lira, Yoon Kim, Chanyeol Choi, Hyeongwoo Kong, Yongjae Lee
Format
A full-day, in-person programme. All sessions are held on site.
01
Opening Remarks and Keynote Addresses (2 x 30 min)
02
03
04
05
Key Dates
Paper Submissions
Submission Opens
August 23, 2026
Submission Deadline
Author Notification
Workshop Day
Workshop
Location
Main Hall, Bocconi University, Milan
Call for Papers
Topics of Interest
Submission Guidelines
There is no length limit or strict format requirement. Formatting according to ACM guidelines is recommended, but we encourage an open format so authors can focus on content. Submissions should be in PDF. The workshop is non-archival and will not have official proceedings.
Additional requirements:
For each submission, at least one author must agree to serve as a reviewer and deliver reviews on time.
The author list cannot be modified after the initial submission, except before printing.
Any changes after acceptance will be communicated directly with the authors.
After submission
Review process
Single-blind (authors known to reviewers, reviewers anonymous). No rebuttal period. All submissions treated with strict confidentiality.
Accepted papers
All accepted papers are invited to the poster session (presenters print and bring their own posters); selected papers may get an oral slot. Non-archival: only author names and titles of accepted papers appear on the website.
Academia
Prof. Yoon Kim
MIT
Prof. Alejandro Lopez-Lira
University of Florida
Prof. Alvaro Cartea
University of Oxford
Prof. Sotirios Sabanis
University of Edinburgh Centre for Investing Innovation
Prof. Yongjae Lee
UNIST
Industry
Jacob Chanyeol Choi
LinqAlpha
Edward Tong
Sina Molavipour
SEB
Alireza Javid
SEB
Didier Rodrigues Lopes
OpenBB
Fayssal El Mofatiche
Flowistic
Jacob Chanyeol Choi
LinqAlpha