A personal quant agent is much closer to reality than it was a few years ago.
I’ve been thinking about whether an AI agent could help manage and improve my existing investment portfolio.
Not by asking an AI which stocks to buy. That’s the easy part, and probably the least interesting one.
The harder problem is building a system that understands an existing portfolio, tracks the underlying businesses, evaluates them against a consistent set of criteria, watches for changes in the investment thesis, and helps decide when to buy, hold, reduce, or exit.
It would need to connect several things that are usually scattered across spreadsheets, financial statements, market data, research notes, and an investor’s own reasoning. It would also need to distinguish a falling stock price from a deteriorating business, and a compelling story from a genuinely attractive investment.
The goal isn’t to automate every decision. It’s to build a system that makes my decisions more consistent, exposes weaknesses in my reasoning, and helps me act when the evidence changes.
I’m starting with the portfolio I already own. The first challenge is to turn my existing investment framework into something an agent can use, test, and improve. From there, I can work towards a system that monitors the portfolio and recommends actions with explicit reasoning.
This is an experiment in what one person can build with today’s AI tools. I’ll document what works, what fails, and where human judgment remains essential.
The question I’m exploring is simple: Can an individual investor build a useful quant system without building a quant firm?