ACI Challenge
In The Coming Wave, Mustafa Suleyman proposed a modern Turing test: an AI that can autonomously turn $100,000 into $1,000,000 in the real world — Artificial Capable Intelligence. We are running that test in public. EventTrader's autonomous strategies work a $100k paper book toward $1M, while the same strategies trade real money in parallel. Every point below is measured, never simulated alpha.
The paper book runs the platform's live autonomous strategies at $100,000 notional — the measured live performance index applied to paper scale (impact-free at current sizes). The live book is real money and is published as an index (base 100 at inception).
The executed paper book trades the liquid Potential Movers basket with $100,000 of paper capital: every rebalance day it buys the top picks equal-weight at the next recorded open, pays a 1% impact haircut on entry and again on exit, never takes more than 2% of a name's trailing daily dollar volume, and sells at the recorded close seven days later. Picks are recorded before their outcome is known; nothing is fitted after the fact. Started 2026-06-26; 12 weeks scored, 3 positive, mean -4.2% per week, last week -6.2%; 2 open positions. The basket rule and its liquidity floor are published on Potential Movers.
The ripple proposal book takes EVERY M&A ripple basket the engine proposed — pending, approved, rejected or expired alike — on $100,000 of paper capital, to measure the strategy without its approval gate. Each proposal is bought at the next recorded open after it was written (never the same session), weighted by the proposal's own legs, sized at the live sleeve's per-deal allocation scaled 12.5x from its $8,000 capital base and never more than a seventh of remaining cash, paying a 1% impact haircut on entry and again on exit and never taking more than 2% of a name's trailing daily dollar volume, then sold at the recorded close eight days later — the live sleeve's own modal hold. Proposals are recorded before their outcome is known; nothing is fitted after the fact. Measured result so far: -4.9% ($100,000 → $95,095) across 103 of the 118 baskets the engine proposed — 97 of which the approval gate rejected or expired, so they never traded live. 93 proposals have closed, 23 of them profitably, mean -1.6% and median -2.4% per basket after costs. 38 positions still open. This line exists to answer one question honestly — is the approval gate protecting the book or costing it? — and the number stands whichever way it falls.
The live book is real capital across the platform's autonomous engines — the Tuatara Hedge Fund sleeves (M&A ripple index baskets, AI index baskets, the crypto long/short hedged pair) and the agent perp desk. It is published as an index (base 100 at inception); dollar figures on this page belong to the paper book only. High variance by nature — the point of a public test is that the number is real either way.