Eligibility
Open to students, researchers, developers, startup teams, and independent builders. Participants must be 18 or older at the time of the event (under-18s may attend with prior arrangement and parental/guardian consent — email contact@prophetarena.co). Organizers, judges, and anyone with access to evaluation internals or non-public benchmark data are not eligible to win prizes.
Teams
Solo participation is allowed; teams of up to 10 are permitted. All team members must be registered. Rosters lock at the submission deadline. A person may not be on more than one team.
Build window & evaluation
Build window: Saturday, May 16, 9:00 AM CT through Sunday, May 17, 5:00 PM CT (~32 hours). Submission deadline: Sunday, May 17, 5:00 PM CT — no late submissions. Evaluation runs continuously for two weeks after the event (May 17 → May 31). Winners announced Monday, June 1, 2026.
Judging
• Forecasting Track: scored on accuracy against the Prophet Arena benchmark. Total score = (team average Brier − market average Brier) × completion rate. A team's total score must be greater than 0 to win or place as runner-up.
• Trading Track: scored on live paper-trading PnL. Standings determined by sum of rankings across metrics (including Sharpe ratio and PnL); lowest combined rank wins. A team's PnL must be positive and the agent must execute at least 14 trades during the evaluation window to win or place as runner-up.
Organizers' and judges' decisions are final.
Prizes
For each track: winning team receives a sponsored trip to South Korea (value up to $2,000) to present at the ICML workshop. Runner-up receives $500. Prizes are awarded per team; teams decide how to split.
Fair play
AI tools (coding assistants, LLM agents) are explicitly allowed and encouraged for building. Participants must not: submit previously built projects as new; manipulate, game, or reverse-engineer the leaderboard or evaluation harness; access non-public benchmark data; train or fine-tune on the specific evaluation questions; violate any provider's terms of service; interfere with other teams' work; or misrepresent authorship. Minimal bug-fix changes during the evaluation window are allowed but discouraged — they can introduce inconsistency and worsen your score. Violations may result in disqualification.
Intellectual property
Teams retain ownership of their work. By submitting, teams grant the organizers a non-exclusive, royalty-free license to review, run, evaluate, and showcase the submission for event purposes. A public GitHub repository is required; MIT or Apache 2.0 recommended. Use of submissions beyond evaluation and showcasing requires separate consent.
Code of Conduct
Prophet Hacks is committed to a harassment-free, inclusive experience for everyone — in person and on Discord — regardless of background or identity. Expected behavior: be respectful, collaborative, and professional. Unacceptable behavior: harassment, discrimination, intimidation, unwelcome attention, hate speech, or disruptive conduct. To report a concern, email contact@prophetarena.co or message an organizer directly on Discord or on-site. Organizers may take any action they deem appropriate, including warnings, removal from the event or Discord, and disqualification, without refund. Applies to all participants, judges, mentors, sponsors, speakers, and organizers.
General
Organizers may update these rules before or during the event; material changes will be announced on Discord. Participation is at your own risk; follow venue (University of Chicago / John Crerar Library) policies and applicable laws. Prizes are subject to applicable taxes and any sponsor- or organizer-specified conditions. Questions: contact@prophetarena.co.