Autonomous Trading Platform

The day job is spent governing enterprise technology portfolios: setting the rules, reading the dashboards, trusting the process to be more disciplined than any one person's gut feeling. This project asks whether the same discipline holds up somewhere with much faster, much less forgiving feedback than a quarterly budget review.

It started as a question more than a plan: could a set of clear, written rules make better trading decisions than the emotions that usually drive them? Fear, greed, and boredom don't touch a Python script the way they touch a person watching a ticker. Building the system was a way to find out, one paper trade at a time, in a domain that has nothing to do with the day job and everything to do with the same instincts for signal, risk, and discipline.

At its core, the platform watches a small, hand-picked list of stocks for a specific pattern: a price drifting far enough from its running volume-weighted average to suggest it's likely to snap back. That's the VWAP mean-reversion strategy at the center of the current build, connected to a live Alpaca paper trading account through the alpaca-py SDK.

The system doesn't just find the signal, it manages the position around it: sizing the trade, watching the risk, and automatically flattening everything at the end of the day so nothing is left exposed to overnight news. The whole thing was built in Visual Studio with Claude Code doing a lot of the heavy lifting on the implementation, while the strategy logic and risk rules stayed a very deliberate, human decision.