Profitable grid trades. A losing account.
Four grid configurations completed profitable cycles, yet every whole account finished negative after unsold inventory was valued.
Trading claims are frozen, measured after real costs and published with their receipts attached, including claims built inside Numerroo itself.
Four grid configurations completed profitable cycles, yet every whole account finished negative after unsold inventory was valued.
A frozen machine-learning classifier was tested on BTC/USD and ETH/USD. Both accounts ended below cash and buy-and-hold.
Five Keel rules were tested on EUR/GBP across two timeframes and two cost treatments. None survived out-of-sample.
Time-series momentum was tested across nine ETFs from 1993 to 2026. Long/short failed; long/flat survived as defensive timing only.
We separate a persuasive headline from the account-level evidence needed to support it.
Inputs, parameters, assumptions, outputs and engine commit are checksummed before publication.
Fees, spread, slippage, inventory and adverse outcomes are counted explicitly.
Favourable assumptions, limitations and the strongest counter-reading stay visible.
The owner gates every claim. Corrections are visible and frozen receipts are never rewritten.
Assay does not sell signals, bots, strategies or broker, bot or course affiliate placements. Negative results are preserved, and a strategy that genuinely survives measured costs is allowed to pass.
Read the boundary and independence rules