Everyone who discovers they have access to a powerful LLM eventually has the same thought: what if I used this for investing?
It’s a natural idea. LLMs can synthesise earnings calls, parse filings, track sentiment across thousands of companies, and do it without the emotional baggage — no FOMO, no anchoring, no panic selling. The pitch writes itself.
Here’s why it doesn’t work, and why you should build it anyway.
The market doesn’t wait for your synthesis
By the time an LLM reads an earnings transcript and forms a view, algorithmic traders have already moved the price. The edge AI gives you isn’t speed — it’s synthesis. But synthesis edges in public markets are thin and temporary. The SPIVA data is brutal: ~90% of professional fund managers underperform their benchmark over a decade. These are people with Bloomberg terminals, research teams, and decades of experience. Your ChatGPT wrapper isn’t the missing ingredient.
Confident hallucination is the exact wrong trait for investing
LLMs construct compelling narratives. That’s the product feature. It’s also the failure mode. An LLM can build you a rigorous-sounding bull case for a terrible stock, and it will feel like analysis. Narrative coherence isn’t truth — and in investing, the difference is your money.
So why build it?
Because the learning is the point, not the returns.
Pick a universe of stocks. Define two or three signals — sentiment from earnings calls, insider transaction patterns, macro indicators. Backtest against historical data. See if anything actually beats buy-and-hold after transaction costs.
You’ll learn more about markets, signal decay, overfitting, and the limits of pattern recognition than any investing course will teach you. And you’ll learn it empirically, not theoretically.
The other payoff is professional. “I built an AI-driven signal pipeline and here’s what I learned about where LLMs add real value” is a better consulting conversation piece than any certification. It demonstrates exactly the kind of judgment clients need: knowing when AI helps and when it doesn’t.
The boring answer is boring because it’s correct
Passive indexing still wins for almost everyone. Low-cost index funds, regular contributions, long time horizon. Buffett’s million-dollar bet against hedge funds proved it at scale.
The real alpha for most knowledge workers isn’t in the market — it’s in career income growth. The expected value of 40 hours of stock research is almost certainly negative compared to 40 hours invested in your actual career.
Build the system. Learn from it. Then put your money in an index fund and go back to work.