Listen to the audio version of this article (generated by AI).
Editor’s Note: Investing is changing faster than most people realize. Markets react in seconds… patterns form and disappear just as quickly… and traditional methods are becoming outdated.
More and more, the edge favors investors who can identify these patterns early — or, even better, employ systems that can detect them automatically.
That’s why we wanted to share today’s guest essay from TradeSmith CEO Keith Kaplan. In it, he explains how his team is using AI to identify specific “signals” in the market — patterns that have historically led to profitable trades — often before they fully play out.
To learn the ins and outs of this system, , where he walks you through how it works and what it’s flagging in the market today.
Take it away, Keith…
“There is an entity that cannot be defeated.”
That’s how Lee Sedol – the world’s greatest Go player – described AI after losing a five-game contest against Google’s AlphaGo AI model in Seoul in March 2016.
Go was invented in China more than 2,500 years ago. Two players place black and white stones on a grid, trying to capture each other’s pieces.
Stones on a Go board (Source: Wikimedia Commons)
It looks easy. But it contains more possible positions than there are atoms in the observable universe. The best players win by feel and intuition rather than brute calculation.
So, Sedol thought he was safe. The game was too creative – too deeply human – for a machine to crack. Before the match, he predicted he’d win every game.
Then came Move 37 in the second game.
Google’s AlphaGo placed a stone in a position no normal player would make – a move so strange that one commentator blurted out, “That’s not a human move.”
He was right.
AlphaGo’s analysis showed that a human player would make that move less than 1 in 10,000 times. But it turned the tide in AlphaGo’s favor, and Sedol lost.
He went on to lose the contest 4–1.
The machine wasn’t smarter. But it could see patterns no human eye could. And that was enough to defeat the world’s best player.
It’s a cautionary tale for investors today. AI models are way better at recognizing patterns now than they were a decade ago. And there’s no domain richer in hidden patterns – or more ruthlessly efficient at rewarding those who spot them and punishing those who miss them – than the stock market.
That’s why my team and I at TradeSmith have spent the last 12 months building a new application of AI for self-directed investors like you.
It uses the same principles that defeated the world’s greatest Go player to spot patterns that flag the day’s most promising trades – 90 minutes before they happen.
I’ll show you how it works today – and pass on the link where you can watch it in action. First, let me give you some background on TradeSmith and what we’re all about.
Modern-Day Prospectors
As TradeSmith’s CEO, I run a team of 65 people, and an annual budget of $8 million, to develop hedge-fund-level analytical systems for self-directed investors.
More than 134,000 people in 86 countries use our software to manage more than $29 billion in assets.
Inside our Research Lab, we’re like modern-day prospectors panning for gold – only we use data and computers, not pans and pickaxes.
We’re constantly testing trading strategies, financial metrics, and data patterns to uncover profitable systems and indicators.
That’s what’s gotten us featured in Forbes, The Wall Street Journal, and The Economist.
Our risk-management software, TradeStops, put us on the map among financial technology firms. It takes the emotions out of investing by showing you the ideal time to sell your stocks.
We’ve also created software that finds hidden seasonality patterns in stocks… spots undervalued options plays… and uses AI to forecast stock moves up to 21 trading days out.
We’re always innovating. But we’ve never gone this deep.
Our new AI-powered system doesn’t look at balance sheets… read earnings reports… or follow news headlines. Instead, it detects tiny anomalies in stocks’ historical data. Then it finds statistical connections between them that a human analyst would never find.
Think of it like a “thumbprint.” Every great trade has one. A unique combination of factors – technical indicators, price patterns, market conditions – that has lined up before.
When those factors align again, our system flags a high-probability setup. Some with historical accuracy rates of 90% or more.
Here’s what happens when we tested it out.
Turning Patterns Into Profits
In January, our system flagged a trade in Qnity Electronics Inc. (Q) – a company I’d never heard of before.
It identified three factors that had aligned only four times over the past decade. Every time those same factors lined up for Q, the stock went up.
On January 28, this signal fired again – and over the next 30 days Q shares jumped 26%.
Or take AI chipmaker Advanced Micro Devices Inc. (抖阴最新版). On a recent signal, our system forecast an 8.4% move in 14 days, based on a pattern with 95% historical accuracy.
The result was an 8.1% gain in 48 hours.
Or take another popular AI stock, Palantir Technologies Inc. (PLTR). Our system signaled a 5.8% move in nine days, again with a historical accuracy rate of 95%. The result was a 15.1% gain in seven days – nearly three times the forecasted return.
These returns are from trading our signals with stocks. Here’s what those same signals produced using options:
- Caterpillar Inc. (CAT): 126% in 72 hours
- Nvidia Corp. (NVDA): 129% in 5 days
- Lockheed Martin Corp. (LMT): 365% in 30 days
- HCA Healthcare Inc. (HCA): 461% in 13 days
- Generac Holdings Inc. (GNRC): 1,082% in 33 days
Now, you’ve seen some of what the signals can do. Let’s look at what’s happening under the hood…
Finding the What, Not the Why
The technology behind this new trading system is similar to what powers ChatGPT and other AI models.
They soak up massive samples of language, find statistical relationships between words, and predict what comes next. Our system uses the same principle – but for numbers.
We fed it data on 2,467 stocks going back 10 years – including interest rates, Treasury yields, Relative Strength Index readings, Bollinger Bands, and intraday trading ranges.
It also looks at indicators most traders have never heard of. Things like:
- Kaufman Efficiency Ratio: It measures how cleanly a stock is trending versus how much it’s just drifting sideways.
- Williams %R: It’s a momentum oscillator that measures where price sits relative to its recent high-low range.
- DMI+/DMI-: They measure the strength of upward versus downward movement independently, rather than just looking at price direction.
No one knows exactly why each signal works so well. Frankly, it doesn’t matter. Like AlphaGo, our AI finds the what… not the why.
It runs each stock through 847 individual calculations daily, compiling more than 2 million trade evaluations every 24 hours.
It’s looking for combinations that have worked before – regardless of whether there’s an obvious reason why.
The result is a trading system that doesn’t care if we’re in a bull or a bear market. It doesn’t need a strong economy or a calm geopolitical environment. It just needs the ingredients to align.
Given what we’ve seen in 2026 – a wipeout in software stocks, oil above $100 a barrel, and rising volatility – that kind of neutrality matters more than ever.
Now You Can Try This Breakthrough for Yourself
This is unlike anything we’ve released in our firm’s 21-year history.
The 30,000 (and counting) signals our system has discovered for 2,467 stocks give you the kind of edge that would otherwise be off-limits for most investors.
Find out more about it by tuning into my .
I’ll walk you through how it works in more detail – including the signals it’s tracking right now and the trades it’s flagging for the weeks ahead.
Keith Kaplan
CEO, TradeSmith
