Inside the Machines That Now Run Wall Street

Listen to the audio version of this article (generated by AI).

Editor’s Note: Some of the most powerful trading firms in the world no longer rely on human instinct.

They rely on pattern recognition — systems that sift through enormous amounts of market data looking for repeatable setups with a statistical edge.

That shift is already reshaping how money is made on Wall Street.

That’s the idea my colleague Keith Kaplan explores in the essay below. He shows how his team built an AI-driven system to uncover repeatable signals across thousands of stocks — and turn them into a simple, actionable strategy.

You can dig into the full framework — and see the system in action — in his Signals presentation, available here.

“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.

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.

From Go Boards to the Stock Market

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.