We built what we wanted to trade with.
MTOPS started as a research project — can a neural network learn the harmonic structure of BTC at 30-second resolution? After a year of iteration, the answer was: sort of, sometimes, and only when paired with a brutal filter chain.
So we stopped calling it a "signal generator" and started building a decision pipeline: eight stages, every one auditable, with dry-run as the default state. The AI's job is to propose. The pipeline's job is to reject almost all of it.
This tool is not financial advice. It's an instrument for traders who already know what they're doing and want their reasoning in writing.