Posted on 08/26/2025 9:33:51 AM PDT by Red Badger
A 17-year-old handed just $100 to ChatGPT and let the AI run a stock portfolio for one month. The results stunned finance watchers, with gains far beyond major Wall Street benchmarks.
===============================================================
In late June 2025, 17-year-old Nathan Smith from rural Oklahoma gave ChatGPT control over a $100 stock portfolio. Four weeks later, the results shocked many: a 23.8% gain—far above the Russell 2000 (+3.9%) and biotech ETF XBI (+3.5%) during the same period. Smith shared his experiment on Reddit, and it quickly spread across tech and finance communities, with coverage from Decrypt, Futurism, and others.
Smith is quick to caution, though: “Nothing about this is financial advice or me trying to sell something, just a small cool experiment I wanted to show.”
How a $100 Portfolio Turned Into a Case Study Nathan Smith’s project started as a response to the constant ads promising AI-powered investment success. Curious, he gave ChatGPT a strict set of rules: it could only buy entire shares of U.S. micro-cap stocks valued under $300 million. He manually executed the trades once a week based on ChatGPT’s picks, while Python scripts using Yahoo Finance tracked performance.
Transparency was a key part of his project. Smith published his setup on GitHub and detailed updates on Substack, allowing others to follow the experiment or replicate it themselves. He emphasized that ChatGPT was not left to run on autopilot—he monitored every step, making small adjustments if the AI contradicted itself or recommended impossible trades.
The AI’s Strategy: Micro-Caps, Stop-Losses, and Discipline The backbone of the experiment was discipline. ChatGPT enforced automatic stop-losses, selling positions if they dropped below a set threshold. It also managed portfolio sizes conservatively, spreading small bets rather than taking oversized risks.
One standout winner was CADL, a micro-cap stock that accounted for nearly half of the total profits. The AI’s decision to exit early locked in gains before the stock reversed—a move that showed a cautious, almost human-like awareness of risk.
By the end of the four weeks, the portfolio had gained nearly $25 on the original $100. Compared to broader U.S. indices, that’s an eye-catching jump in such a short timeframe.
Sharpe and Sortino Ratios Tell the Story Smith didn’t stop at raw returns. He calculated professional risk metrics to see if ChatGPT’s success held up under scrutiny.
Sharpe ratio: 0.94 — strong, though just shy of the 1.0 mark typically viewed as “very good.” Sortino ratio: 2.00 — excellent, showing high returns with relatively little downside volatility. These numbers suggest the AI wasn’t just gambling—it was producing returns with calculated risk, at least over the short window.
Still, one month of trading isn’t enough to claim reliability. Academic research supports that caution. A University of Florida study earlier this year found that AI-generated stock picks often outperform on paper, but the advantage fades quickly when scaled up or used in real markets.
Ping!...................
The same way the Pelosi clan got so wealthy right?
Without that one lucky pick and exit, it wouldn’t be anything to write about. It was most of the gain.
He sold his Soul for 23.8 percent...
Four weeks is hardly a test.
“Without that one lucky pick and exit, it wouldn’t be anything to write about.”
It wasn’t luck.
“It was most of the gain.”
Half his gain. Even without that gain it significantly outperformed the market.
Having an edge in anything only helps so long as you’re the only one who possesses that edge.
If AI is so good at managing a portfolio, then why do all of these AI startups need VC funding ?
From late April to late July, it wasn’t at all difficult to beat the indexes, even by the amount he did with AI. He should have done his experiment starting in mid-February and watched how it did until about early April when just about everything was tanking. If he made the same sort of gains during that period, I’d pay more attention.
Monkeys typing Shakespeare.
-PJ
If a million people invest $100, could ChatGPT move the markets? And they==is, subsequently improve performance on an ongoing basis?
how long until there’s a ChatGPT investment portfolio ETF?
And it so happens that in the six month following what’s known as the golden cross that occurred in early july when the big beautiful bill passed— that golden cross has been followed by stock market gains of about 30%. Plus the stock market gains another 10 percent or so in the next 18 months.
I’m of the opinion that with the vast investments coming into the USA and the vast technological transformation that’s coming to radically change US manufacturing and just about everything else—the USA is on a multi decade bull market comperable to the bull market of the late 1800’s with the industrial revolution.
I spent several hours over a period of weeks researching where to put some money that was under employed. After I picked four stocks to buy in differeing proportions I made a very simple query to whatever AI Google uses. Out of the four stocks I picked it recommended three. The fourth one, not recommended, was an outlier I bought little of on a speculative action not within the simple goals I gave the AI query.
The other interesting thing is that these three appeared as long term investments recommended by analysts. One of them was highly recommended as a darling. That was during the April bust. Things went well until the highly recommended company hiccuped just a little on quarterly earnings and then the analysts absolutely pilloried it, even the one that recommended it as a long term good buy with glowing prospects within a year.
I worked at a very large multi-national oil company in business development for a time. I watched the analysts opinions closely and the actions after conference calls. To see them in action one would think these people were setting themselves up for self-filling longs and shorts.
Over time and because of my work experience I also learned that blue chip companies and dividend aristocrat companies will empty the building and even sell it to keep paying a relatively level dividend, they will even borrow money to do it.
After all my time doing and watching the last 50 years, the only thing I really regret that I wish I had known and done from the beginning would have been to invest my first, last and every other dime in an S&P ETF. I think I am in good company by such an observation as Warren Buffet suggest the very same thing. My slight recompense is that such a thing did not exist in it’s form today when I began investing. If I had done that I would be better off than I am by at least 50% with a whole lot less wear and tear. What is more, I would be living strictly on the 1.2% dividend and never sell a thing. It is so simple a child could do it. If the shares go down 25% you may not like it but you aren’t selling and in all likelihood they will recover by 33% in three years or less, that is unless obama and bidet are in power then it takes 7 or more years to claw back.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.