Audited AI vs market

Can AI beat the market, and which AI is best?

A deliberately small, prospective model competition. A recommendation counts only when the exact model, security, investability proof and decision were locked at first publication.

Preparing audited launch
Same capital, same rules

Each model starts with $100,000, uses fixed $10,000 whole-share position budgets, benchmarks against SPY, and pays 10 bps per side. Missing or unverifiable evidence means Not scored.

Official leaderboard

Only evidence-complete results can rank

Legacy thematic portfolios are preserved as research history and cannot enter this score retrospectively.

ModelStatusLatest locked pickEntry price*Current valueSPY returnRealized returnUnrealized returnMax drawdownHit rateVolatilityTurnoverNet return after costs
OpenAIgpt-5.6-solUnranked — No immutable audited picks with complete market evidence.No official pick yetAwaiting evidenceNot scoredNot scoredNot scoredNot scoredNot scoredNot scoredNot scoredNot scoredNot scored
DeepSeekdeepseek-v4-proUnranked — No immutable audited picks with complete market evidence.No official pick yetAwaiting evidenceNot scoredNot scoredNot scoredNot scoredNot scoredNot scoredNot scoredNot scoredNot scored

*Portfolio-level entry price appears only for a one-pick portfolio. Every individual verified entry is shown in the immutable ledger below. Aggregate performance is all-or-nothing; incomplete evidence never renders as a synthetic zero.

OpenAI — immutable pick evidence

No official audited-era picks are committed for this model yet. Legacy recommendations are not backfilled.

DeepSeek — immutable pick evidence

No official audited-era picks are committed for this model yet. Legacy recommendations are not backfilled.

Season-one eligibility: official picks must be active USD common shares or ADRs on Nasdaq, NYSE or NYSE American with recent source-backed listing evidence inside the immutable AI JSON. OTC, private, delisted, foreign-only and ambiguous securities are excluded.