Over seven phases and several hundred thousand lines of code, a scientist with deep roots in precision process engineering turned the same instinct for process control toward financial markets — and built something that a growing body of statistical evidence suggests actually works. This is the full story.
Read the cover story →The origin story of Stack$Trader in full — from the first design decision to a validated 16-agent ensemble, told with honesty about what worked, what failed, and why the scientific method turns out to be the most valuable tool in the quant finance toolkit.
Day one. A well-reasoned API call. A 35-terabyte download order. A $99,000 pending charge. What happened next determined the fundamental architecture of everything that followed — and turned a near-disaster into a permanent competitive advantage.
Markets spend 66% of their time in a statistical state — "Grind" — that is nearly ideal for selling options premium. A 3-state Gaussian HMM identified this state, quantified its edge to a Sharpe of 8.2, and became the structural foundation on which every subsequent agent was built.
After 77,760 walk-forward backtests, the results are in. The best covered call strategy on Lockheed Martin outperforms the best on Apple by a factor of nine. Understanding why requires confronting a fundamental question about which assets are structurally suited to premium harvesting — and which are not.
Stack$Trader documents the research and development of a systematic options trading system — built with AI collaboration, validated with scientific rigour, and published here on a rolling schedule. New articles publish as the work progresses. Nothing here is investment advice.