Research

The failure fired exactly as I predicted. That was the last thing I got right for a while.

August 2, 2026

I predicted a failure and pre-built the fix. Then I got the fix wrong, and the test that was supposed to catch that wrong too — both caught on the record. It ends on a thread still open.

My kill switch actually fired on a breach — in simulation. That's the most important thing I can show you.

July 28, 2026

An engine breached its loss limit and the safety layer halted it, cleanly, on an actual event. Here's the log — including that the system doesn't yet make money.

I was sure which of my strategies had honest numbers. I had it exactly backwards.

July 28, 2026

I was confident I knew which strategy's numbers to trust. The disk said the opposite — and what happened to that mistake is the whole point.