There is something about today’s hustle culture that pushes young men to combine crypto, gambling, AI and day trading into one ungodly pot of get-rich-quick hopelessness. This is likely driven by desperation: when the system’s failed you, chaos seems like a viable option to change the status quo.
And so, if you’re looking to make a quick buck on the stock market (trading degen options), you might be tempted to use AI to find out which stock will go up or down next. After all, it’s pretty much a god? Spoiler alert: it’s not. But since we’re here, let me explain how you could, theoretically, misuse AI for stock price prediction.
I stress this now, and I will stress it again : this is a terrible idea.
Hey, is anyone already doing this?
When I say AI, of course, I mean something like ChatGPT — a large language model — that any schmuck like you and me could use. Not some sophisticated quant model that actually resembles true AI research. Those are for the Real Gamblers.
Renaissance Technologies (pretty well known hedge funds — if you don’t know them them, please put your money in the S&P500 and stop reading), for example, has built proprietary algorithms that give them an edge in stock prediction. Their Medallion fund, now mostly run for employees, returned 66% annualised before fees over a 30-year span from 1988 to 2018.
They don’t use ChatGPT; they have what we might call a real predictive model. For the average Retail Investor, AI often means a chatbot, a trend-surfing algorithm wrapped in the illusion of intelligence.
Tel Aviv-based Bridgewise is one of the companies trying to package this illusion into a marketable product. They’re launching a chatbot named Bridget, approved by the Israel Securities Authority, to offer recommendations on which stocks to buy and sell. They’re even teaming up with one of the country’s biggest banks, Israel Discount Bank, to roll it out.
Four ways to use ChatGPT to make money on the stock market
Let me make one thing perfectly clear (again) before we go any further: You shouldn’t use AI for stock prediction. The risks involved in trying to outsmart the market are significant, and chances of success are slim.
In fact, you probably shouldn’t even be picking individual stocks at all. According to research done by Nobel Prize-winning economist Eugene Fama, markets are largely efficient, meaning that all available information is in theory already reflected in stock prices. GME was an anomaly. BTC is an anomaly. There is no short squeeze coming. Save yourself the drama — just put your money in an index fund and call it a day. Index funds are diversified, low-cost, and tend to perform well over the long term, with the S&P 500 providing an average annual return of around 10% over the past century.
But let’s say, despite my warning, you’re dead set on using AI for this. I see a few approaches that aren’t entirely ridiculous.
Analyze annual reports and spot trends
You could collect a company or an industry’s latest annual reports, earning calls readouts and news / interviews, then feed that to an LLM (ChatGPT Premium lets you do that with PDFs).
Then, you could combine that data with disruptive trends you have an eye on (PESTLE is your friend there). GDP growth, unemployment rates, interest rate changes... all are valid and will have an impact.
Asking ChatGPT which companies are more likely to be impacted (positively and negatively) by those trends — and why — is likely to yield interesting results. Those trends were probably mentioned across an industry if they matter; doing research this way could also help you identify trends you may have missed in your earlier analyses. AI can quickly summarize the flood of corporate language into something a bit more digestible… and actionable.
Assess sentiment in corporate communications
If you’re curious about the mood within a company’s official communications (or less official interviews), you could ask ChatGPT to provide a sentiment analysis. The tone of communications can sometimes give you a hint of what’s going on behind the scenes — optimism, caution, defensiveness…
For instance, a cautious tone might indicate upcoming challenges, while an overly optimistic tone could suggest an attempt to mask underlying issues.
According to a 2019 study published in the Journal of Financial Economics, sentiment analysis of corporate disclosures can predict stock returns in the short term, making this approach worth considering. In fact, hedge funds are increasingly using NLP to gauge market sentiment and adjust their strategies accordingly. Which is why tools like AlphaSense and Dataminr are making a killing. But hey, you don’t have access to those tools.
Ask for numerical help / insights
AI can simplify financial concepts. If you don’t know what WACC is for example — and why it’s important — you can ask ChatGPT to explain it. Or even make it do some back-of-the-envelope calculations / comparisons for you.
This can help you get a general understanding of what’s going on if you’re not a finance expert. Key metrics like return on equity, earnings per share, and debt-to-equity ratio are also useful to analyze, and AI can help break these down in simple terms.
This would improve one’s financial literacy and lead to better choices / capital allocation. Some of those better choices would be akin to putting money in the S&P 500 and quitting day-trading and settling down with that nice girl from your home-town. But who am I to judge…
Draw on historical data
When you see certain market trends developing, you could ask an algorithm to give you historical context. How did similar scenarios play out in the past? What were the drivers? What happened afterward? This could give you an idea of what may happen in the future. After all, those who don’t know history are destined to repeat it.
It’s important to note, however, that historical trends are notoriously imperfect predictors. Financial markets are influenced by countless variables, and no pattern guarantees future outcomes. But, understanding past trends can provide valuable (if not actionable) insights, particularly when combined with other data points.
Again, I want to emphasize: you probably shouldn’t do this. AI has a lot of use cases, but making you rich through stock prediction is not one of them.
The folks getting rich off AI aren’t predicting stock prices; they’re selling AI products or developing high-frequency trading algorithms that you don’t have access to. Risks include misinterpretation of data, over-reliance on incomplete information, and market volatility that no AI can fully predict. So unless you’re willing to risk it all for the thrill — in which case, just go to a casino — keep your money where the odds are less stacked against you.
Good luck out there.
Comments