Have you ever wondered what would happen if you let an AI handle your money in the stock market? I thought about it for weeks. Everywhere I looked, I saw ads saying that AI can find the next hot stock. Most of those ads looked like scams, so I always ignored them. But the idea stayed in my mind. Could AI really do it? Could it actually pick better stocks than humans?
So I decided to try something bold. I gave ChatGPT one hundred dollars and told it to pick trades for me. I wanted real results. No fake screenshots. No pretend backtests. Just a simple test with real money that I could afford to lose. Two months later, the results surprised me more than I thought they would.
Where This Idea Came From?
This test was inspired by a project on GitHub created by Nathan B. Smith. He wanted to see if ChatGPT could beat the market by picking tiny companies, also called micro-cap stocks. These are small businesses worth less than 300 million dollars.
Nathan started the project in June 2025. The plan was simple. ChatGPT would decide which stocks to buy and sell. Nathan would place the trades. And everything would be tracked openly on GitHub. The goal was to see if the AI could beat the S and P 500 and the Russell 2000 by December 2025.
How the System Works?
The process is easier than you may think.
- Nathan collects stock prices and market data every trading day.
- He sends the data to ChatGPT and asks which micro-cap stocks to buy or sell.
- On weekends, ChatGPT reviews the full portfolio and looks for new picks.
- Nathan makes the trades in a real account. He uses rules like “sell if it drops 10 percent” to reduce risk.
- All the results are saved in CSV files and displayed in charts.
It is like having a smart friend check your piggy bank every day and tell you what to do with it.
Tools You Need
The project does not require expensive gear. It runs on simple tools you can install for free.
- Python
- Pandas and yFinance to collect stock prices
- Matplotlib to draw charts
- ChatGPT to pick stocks
All you need is a computer, the internet, and some storage for CSV files.
How It Performed?
By late August 2025, the portfolio was up almost 30 percent. The S and P 500 was up only 4 percent. Even the Russell 2000 could not keep up. Some weeks looked like a rocket launch.
But remember, micro-cap stocks are risky. One bad pick can wipe out gains quickly. That is why everything is tracked with logs and charts to keep it real.
The Results After Two Months
When I checked the numbers after two months, here is what I found.
ChatGPT’s portfolio had grown by 29.22 percent. The S and P 500 index only grew by 4.11 percent in the same time.
That does not mean ChatGPT is the next Warren Buffett. But it also did not crash or lose everything. The little experiment worked better than I expected.
Can You Try It Too?
If you are curious, you can try it yourself. The GitHub project has a “Start Your Own” folder with templates. You can:
- Install the needed tools.
- Run the script called trading_script.py.
- Give ChatGPT your portfolio numbers.
- Let it suggest trades.
- Use the Generate_Graph.py file to see your results.
Everything is open and ready for anyone who wants to play around with it.
Conclusion
I started this test just to see if ChatGPT could handle a small stock portfolio. After two months, the results were better than I ever thought. From one hundred dollars, it managed to grow almost thirty percent while the S and P 500 only grew four percent.
Does this mean you should let AI run all your investments? Not at all. Stocks are risky, and micro-caps can be wild. But this little experiment shows one thing clearly. AI is no longer just writing emails or editing photos. It can also take real money into the market and sometimes even beat it.
This was the story of Civil Learning. Here’s a link to connect with him if you’re interested.
