Experiment: asking AI to value stocks like Buffett and Graham
October 7, 2026 · by Iftikar Please note: this post is strictly educational — a learning experiment of my own. Nothing here is financial advice or a recommendation to buy, sell, or hold any security. I am not a financial advisor, and AI-generated analysis can be wrong. Do your own research and talk to a licensed professional before investing.
Recently I started experimenting with an AI agent for stock analysis — not to get stock tips, but to learn how great investors think.
What I did
I picked a stock and gave it to the AI agent with a specific instruction: analyze this company the way Warren Buffett would. Estimate the intrinsic value of the business — what it is actually worth based on its earnings power — and tell me whether the current price looks overpriced or underpriced compared to that value. And crucially: explain every term properly along the way, so the analysis itself becomes a lesson.
One ground rule before any of that: the stock had to be Shariah-compliant first. For a Muslim investor the order is fixed — halal screen first, valuation second — because there's no point finding a wonderful underpriced business you can't own. I ran the compliance check with screening data like the kind I covered in my Zoya API deep-dive, and only then asked what Buffett would pay for it.
Then I ran the same exercise through a second persona: Benjamin Graham, Buffett's teacher and the father of value investing — the quantitative, no-nonsense lens of Mr. Market, net-nets, and demanding a margin of safety before paying for anything.
Why personas?
Because the point was never the verdict on one stock. The point was the method:
- Buffett's lens asks: is this a wonderful business at a fair price? What is its intrinsic value — the cash it can realistically generate over its lifetime, discounted back to today?
- Graham's lens asks: where is the margin of safety? Am I paying far less than the business is worth, so that even if I'm wrong, I'm protected?
Learning to ask those two questions — in the right order, with the right terminology — is learning to think about investing smartly. That is the skill I want to build, and to share.
What the agent actually did
For each stock, the agent walked through the financials step by step: revenue and earnings trends, free cash flow, debt levels, valuation ratios — and translated each one into plain language. Every piece of jargon got a proper explanation, because an analysis you can't understand is an analysis you can't learn from.
After the first runs, I started applying the same approach to a few more stock profiles, comparing how the two personas weighed the same company differently. Those write-ups will appear here as the experiment continues.
Where this is going
This is experimental, and I'm learning in public. The goal is to show — practically, hands-on — how an ordinary person can use AI to study investing the way the masters do it: carefully, quantitatively, and with humility about what you don't know.
If you try this yourself, start by asking questions: What is this business worth? Why? What could I be missing?
Please read: this experiment is strictly educational. I have no intention of promoting any stock or investment — any companies mentioned are examples for learning, not endorsements. I am not a financial advisor, this is not financial advice, and AI analysis frequently makes mistakes. Past performance never guarantees future results. Do your own research, and talk to a licensed professional before investing.