01
AI-Assisted Research on an Options-Market System
A rules-based options strategy, treated as a research problem rather than a bet.
- Role
- Independent · self-directed
- Tools
- Claude (research & development partner), live market data, a VPS for continuous forward-testing, Python.
- Status
- Ongoing — in continuous forward-testing.
I wanted to understand whether a rules-based options strategy could hold up outside of theory, so I treated it like a research problem rather than a bet. Working with Claude as a thinking and development partner, I studied the underlying market structure, translated ideas into testable rules, and backtested them against historical data to find where they broke.
From there I moved to forward-testing: running the system against live market data on a VPS so it could be evaluated continuously and honestly, without the hindsight bias that makes backtests look better than they are. A lot of the real work was discipline — resisting the urge to act ahead of what the data had actually shown, and letting the system prove or disprove itself.
The point wasn't a single result. It was building a repeatable, evidence-first way of testing an idea — and learning to trust a process over impulse.
