An autonomous quant research programme against Binance USD-M perpetual futures, handed over whole — including the part where it failed.

Its own verdict is on the tin:

profitable_strategy_demonstrated = false

No clean-holdout result. No live profitability. No order was ever sent. That sentence is the product, and everything here is arranged so you can check it.

INSIDE
• ~58,000 lines of Python: research engine, 14 clean-slate mechanisms, shadow execution runtime
• 69 predeclared freeze contracts, each committed and hashed before its first number
• 312 evidence artifacts and a hash-chained forward decision ledger
• Cost and capacity measured per symbol and per size, not assumed
• 419 tests passing offline from a clean extraction
• 60-page research vault, 21-chapter guide, 7 worked case studies

ONE MINUTE, FULLY OFFLINE
python quickstart/first_run.py

No account, no API key, no network, no market data. It reproduces the recorded result that 128 child orders were sendable at $100,000 of modelled capital and 0 of 128 at the account’s real balance.

NOT a bot, a signal service or investment advice. Market data is not included. The order path was audited by the project itself as not production-safe. Closed archive: no updates.

FOR quant researchers and infrastructure engineers who want research governance and an execution design with an honest post-mortem.

NOT FOR anyone who wants a strategy to run tomorrow. There isn’t one, and the archive says so on every page.