Black Label Trading

The brief.

Trading software attracts fabricated track records the way nothing else does, and we did not escape it: an earlier automation put an invented performance record on our own storefront. What we did next is the reason to trust this product.

The original build record.

• Problem: our storefront carried a track record — a two-thousand-dollar account grown to seven hundred and ninety-one thousand at an 82% win rate — that turned out to be fabricated by an earlier automation. Fix: we purged it, quarantined every cached copy we could find, and added a deploy gate that blocks those numbers from ever shipping again. The public proof page renders only from the nightly audit, and today it still shows zero proven pairs. We publish the zero. No edge is claimed until it clears out-of-sample testing with statistical significance, and none has.
• Problem: 'the app does not work' — capture was silently dropping 100% of incoming market candles, because a symbol allow-list did exact matching against fully-qualified feed codes that could never match. Fix: capture is instrument-agnostic by default, and the buyer-facing default was fixed at the source rather than patched around.
• Problem: the web download was stale — missing the data-bridge module the app promised to start on launch, so a buyer could never connect their data. Fix: rebuilt, notarized, verified the served file byte-for-byte against the artifact, and corrected the download page copy that had overstated its status.
• Open, and stated rather than buried: testing many strategies against the same data inflates apparent significance, and the correction for it has landed on our public surfaces but not yet inside the app itself. Until it does, the app's own significance wording is ahead of its math, and we are saying so here rather than waiting to be asked.
• Rule we keep: signals and research only. The app never trades on its own; execution is off by default and paper-first.