Building My Investor Operating System

I have always been fascinated by the world of investing - from quantitative finance and investment banking to mutual funds, venture capital, and private equity. But I eventually realised that pursuing an MBA in finance was not the path I wanted to take just to explore these fields more deeply.

With AI becoming more capable, exploring these fields has never been more accessible. It is now possible to research ideas, analyse companies, test investment frameworks, and learn by building.

Most of us know that investing deserves regular attention, but actively researching companies and managing a portfolio can take significant time. For someone with a full-time role and other responsibilities, maintaining that level of consistency is difficult.

I wanted to find a practical way to move closer to something I have always been curious about, without turning it into another full-time commitment. That led me to this experiment.

So I decided to create my own research desk: a team of specialised AI agents that studies markets, analyses companies, evaluates risks, challenges conclusions, and recreates parts of the research process used by investment firms.

Building the Team

Professional investment decisions are rarely based on the work of one person. In an active investment management firm, research analysts study companies and financial statements, quantitative analysts examine data and portfolio behaviour, risk teams monitor potential losses and exposures, and portfolio managers decide how capital should be allocated.

The exact structure varies across mutual funds, hedge funds, quantitative firms, venture capital firms, and private equity firms. My system most closely resembles a simplified public markets research desk, while borrowing selected ideas from the wider investment industry. I wanted my research process to follow a similar division of responsibility.

Putting the Research Desk to Work

Once the team was in place, I ran the complete research process across the Indian and US markets. The system shortlisted eight companies, analysed them across fundamentals, market timing, risk, and quantitative factors, then produced a preliminary ranking. The lead candidate was reviewed by a research verifier and examined by a thesis challenger whose role was to identify weaknesses, questionable assumptions, and contradictory evidence.

What the First Run Produced

The first complete run shortlisted eight companies across the Indian and US markets. After verification and challenge, the system revised some of its initial conclusions. Mahindra and Mahindra was moved from Research Buy to Watch, while ICICI Bank and Wells Fargo remained the strongest research candidates. UnitedHealth received an Avoid rating after the risk officer raised unresolved regulatory concerns.

The most valuable outcome was not the final ranking. It was seeing the system question its own analysis and change its view when the supporting evidence became weaker.

You can read the full research note here.

What Comes Next

This first run showed that the system can research companies, compare ideas, verify claims, and revise its own conclusions. What it has not proved yet is whether those conclusions can lead to better investment outcomes.

I am starting with ICICI Bank as the first live experiment. I will track the original thesis, quarterly performance, identified risks, and returns against a relevant benchmark.