This AI Finds Patterns within the Market’s Chaos – With Big Chances to Profit

By Mike Burnick September 17, 2026

Editor’s Note: Good evening – Mike Burnick here.

The biggest market moves reward whoever sees them first. A stock that jumps on earnings, a sector that turns before the headlines catch up – the money gets made by the people positioned ahead of the move, not the ones reacting to it.

For decades, that kind of edge belonged to the biggest players. The top quant funds spent fortunes building machines to find the patterns hidden inside it, while everyday investors never had a shot.

TradeSmith CEO Keith Kaplan set out to change that. His team built a different kind of AI – not a chatbot, but a model trained to find the patterns buried in the market’s turbulence and project where any of 2,334 stocks is likely to head next – to the penny, with 85% backtested accuracy.

Below, Keith explains how it works – and how you can put it to work yourself.

Now, here’s Keith…

Richard Feynman could explain almost anything in the physical world — except what was happening in his coffee cup.

He won a Nobel Prize for working out how light and electrons behave, in equations so exact their predictions matched reality to more than ten decimal places.

And when he sat on the commission investigating the 1986 Space Shuttle Challenger disaster, he made the cause impossible to ignore. On live TV, he clamped a piece of the shuttle’s rubber sealing ring, dropped it in a glass of ice water, and showed the room that it had gone stiff.

But one problem beat him his entire life. It wasn’t in the physics of light, or the wreckage of a space shuttle. It was in the cream curling into his coffee. In the smoke rising off a match. In the water sputtering from his tap.

It was turbulence — the instant a smooth, calm flow trips over itself and erupts into chaos.

One piece of that puzzle had stumped mathematicians since the 1930s. It sounds simple enough: Can a smooth, calmly flowing fluid suddenly break down on its own? But for 90 years, no one could prove it either way.

The question was so hard that, in 2000, a group of mathematicians named it one of the seven greatest unsolved problems in all of math — the Millennium Prize Problems — and put up a $1 million reward for anyone who could crack it.

Then, early this month, it happened. But it wasn’t a human mathematician who cracked it. It was a swarm of 10,000 AI agents

Those agents were built by OpenAI, the company behind ChatGPT. And what they pulled off — finding a pattern no human had ever been able to see — isn’t only useful for solving a 90-year-old math problem. It turns out to be just as useful in the stock market.

That’s why my team and I have spent years of work – and millions of research dollars – putting the power of AI in the hands of ordinary investors.

Not a chatbot. A different kind of AI — one trained to find the patterns hidden in the market’s own chaos, and to project where any of 2,334 stocks is likely to head over the next 21 trading days.

And it doesn’t act on every forecast — only those that have proven right in our testing at least 85% of the time.

Tools like this used to be locked inside the world’s top hedge funds. The firms that had them didn’t sell them and didn’t talk about them. Now, we’re bringing that same edge to Main Street.

In testing, a simple five-stock version of it returned an average annual gain of 239% — across seven years that ran straight through the pandemic, the 2022 crash, the 2025 tariff selloff, rising interest rates, and two wars.

On Tuesday, Sept. 22, at 10 a.m. Eastern, I’ll show you how it works and how you can use it in your own trading. You can RSVP for that here.

Today, I want to show you the idea behind it — the same idea those AI agents used to crack a 90-year-old math problem, now put to work on the stock market.

700,000 Forecasts a Month

Think about how ChatGPT works.

It’s read so much writing that it’s learned how words tend to follow one another. Now picture that same technology aimed at a different target.

Instead of language, AI models can be trained on numbers — and there’s no more potentially lucrative set of numbers than the stock market.

Every price, every gap, every sudden spike in how a stock moves. Train an AI on that kind of data, and it learns how stock prices tend to behave — enough to project where a stock is likely to head next.

It doesn’t know what a company does. It doesn’t read the news or listen to tips on CNBC. But it recognizes patterns in large sets of numbers.

Show it how a stock is trading today, and it finds moments in history that look like this one — then forecasts where the price is likely to go next based on what has happened in the past.

We call it Predictive Alpha. Every trading day, it forecasts prices on 2,334 stocks — more than 700,000 projections a month.

We’ve instructed the model to stay quiet on most stocks. It only flags one when that stock has hit its forecast at least 85% of the time in the past. In a typical week, fewer than one in a hundred clear that bar.

Here’s what those forecasts have looked like.

On July 27, 2023, the model projected Opendoor Technologies (OPEN) would reach $4.87. The stock hit that exact price 24 hours later — a move of 9.4%.

On June 16, 2025, it pointed to Garmin (GRMN) trading at $212.83. Garmin got there to the penny, and faster than expected — a 4.5% move, in 15 days.

In May 2025, it projected Tesla (TSLA) would reach $302.89 within 21 trading days. Even with tariff uncertainty hanging over carmakers, the stock hit the exact figure — a 5.2% move, in just one day.

It’s not a crystal ball. Markets shift. New patterns emerge. But hidden in the seeming chaos are patterns that repeat — if you know how to look.

This AI Reads the Market’s Turbulence

Every trading day, the market takes thousands of small shocks — a Fed comment, an earnings miss, a factory delay a world away.

Most of the time, it absorbs them and moves on. Then one of them tips the whole system, and a calm market turns into a crash.

That’s a type of turbulence — the same problem that stumped mathematicians for 90 years, until AI cracked it this month.

Markets run on the same kind of chaos. On any given day, billions of data points are changing by the second — interest rates, earnings, currencies, commodities, and far more. Move any one of them a little, and the effect ripples out across thousands of stocks. It’s far more than any person could ever track.

But it’s exactly what this new kind of AI is built to read.

Wall Street already knows it. The top quant funds — Renaissance Technologies, Two Sigma, Citadel — have run on machine learning for years. Now the rest of the Street is scrambling to catch up, in what’s become a full-blown AI arms race.

Morgan Stanley recently warned that a wave of “creative destruction” is coming for investing itself — and firms that don’t adapt risk being left behind.

Very few Main Street investors have anything close — yet. That’s what we’re setting out to change.

Join me on Tuesday, Sept. 22, at 10 a.m. Eastern. I’ll walk you through the technology behind our AI system and the forecasts it’s flagged. I’ll explain why the five-stock strategy works — and how you can use it to build an AI-powered portfolio designed to capture the market’s biggest short-term moves.

I’ll even pass along one of the system’s top trades, as a thank-you for joining.

Here’s that link again to RSVP. I hope to see you there!

Sincerely,
Keith Kaplan
CEO, TradeSmith