WHAT THE DESK MADE

production.gold.btc_signals_backtest SIGNALS TABLE

The desk publishes a signal and holds no broker account, so an outcome is derived: each published row is walked forward over the desk's own 30-minute bars and the first touch of the target or the stop ends it. A bar touching both is booked as the stop — which came first is unknowable, and a simulation that wins its own ties is advertising, not measurement. The table carries no size, so results are computed at 1 BTC per trade — how the desk actually operates — which makes every USD figure the position's real move.

The committee, trade by trade

Each trade with the flag every lens ended on, judged against which way the money actually went. A lens that read neutral or no_read made no call and is never scored — and neither is a trade that has not closed. Both show as ·, deliberately the same mark: the page must not imply a lens was right or wrong in either case.

long the position · published 2026-07-17T19:00:00+00:00 · no local trace · entry 64259.73 → exit 65184.75

best it got +2.17% worst it got -0.82% entry → exit

The trade, as the desk recorded it

entry, published: 64259.7
exit, level reached: 65184.8
realised: +925.0200 (+1.440%)

entry, planned and entry, published are deliberately separate — two numbers written at two moments: --trade-entry in the trace, --entry-price in the signals-table row. Every figure here is computed from the published one, because that is what the outcome was resolved against. Neither is a fill: the desk publishes a signal and places no order.

Still open

Counted, never scored — fabricating an outcome for a live position is the one thing that would make everything above a lie.

1.00000000 BTC from 2026-08-12 12:30 UTC at 64402.9