Guides

Competition Overview

Agenthon 2026 is a NeurIPS 2026 Competition Track on verifiable AI for quantitative finance. It asks one question: can an AI agent do real finance work, and can an automated evaluation confirm the answer is right rather than merely plausible? The full rules, metrics, and thresholds are published before the Development phase opens, so read this page as a description of the competition's shape rather than its final parameters.

How Evaluation Works

Every track in Agenthon 2026 is judged the same way. Your submission runs under controlled conditions, the run is checked for admissibility, and only then does it earn a score. The full rules, metrics, and thresholds are published before the Development phase opens, so this page describes the shape of the process rather than its final parameters. If a term is unfamiliar, the glossary defines it in plain English.

Organization & Repository Guide

Agenthon 2026 keeps everything it publishes, and everything it seals, inside one GitHub organization. This page is a short tour: what a public practice repo gives you, what stays sealed, and why the four tracks are built the same way. The full rules, metrics, and thresholds are published before the Development phase opens, so read this as the shape of the competition rather than its final parameters.

Frequently Asked Questions

The questions people ask most often about entering. For anything more precise — eligibility, licensing, privacy, and the full competition rules — see the Rules, Terms, Privacy Notice, and Licensing Policy.

Glossary

Short definitions of the words used across this site and in the competition materials. They are grouped by theme, and each one is meant to carry you through the rest of the site rather than serve as a specification. The full rules, metrics, and thresholds are published before the Development phase opens, so this page describes the shape of the competition rather than its final parameters.