Creator of DeepStack poker A.I. launches stock prediction startup

Jon Pill
Posted on: May 28, 2022 09:00 PDT

Equilibre Technologies is a new start-up that hopes to use game theory and reinforcement learning to build a computer that can beat the stock market.

The company was founded by Martin Schmid, Rudolf Kadlec, and Matej Moravcik who left DeepMind to start Equilibre. DeepMind is owned by Alphabet Inc. — the company that owns Google — and has been behind many of the biggest advances in the field of A.I.

The three founders have a long history in artificial intelligence and were instrumental in bringing DeepStack into being in 2017. DeepStack was the first A.I. to seriously challenge human players at Texas hold'em.

The DeepStack project followed on from the success of AlphaGo, extending the game playing A.I. from games of complete information (like Go and chess) to those of incomplete information (like poker). Now the team at Equilibre is going to try and move from the simple rules of card games to the chaos of the stock market.

“Our idea is that rather than playing poker, our algorithms will play algorithmic trading,” Schmid told CNBC in a recent interview. “We are also looking into crypto[...]. Most of the trading out there is already algorithmic. We just want to create better algorithms than the ones that are already out there.”

Schmid, Kadlec, and Moravcik were also part of the team that built Player of Games, a powerful A.I. that could learn to play almost any game (including chess, Go, and poker) proficiently. So they are well equipped to tackle the challenges of building an A.I. of this sort.

What Equilibre aims to do

The Equilibre website is oddly sparse for a tech company. The text on the front page consists of four bullet points boasting about the company's "world class advisory board" and the qualifications of its main team. It also lays out the company's goal: "EquiLibre is using game theory and reinforcement learning to build the next generation of algorithmic trading."

Given the team involved, we can expect the programming to focus on building an A.I. that can be programmed with the rules for playing the stock market and then set off to learn how to beat the game by itself.

This kind of self-learning is what allowed AlphaGo to improve so rapidly that it could beat every human player in the world after just a few days of studying the game.

Equilibre has been raising funds from venture capitalists across the world, with its largest seed round coming from the Czech Republic. The amount remains undisclosed but is likely to be significant given the pedigree of the team and the potential upside.

It will be interesting to see if the stock market gives way to a similar kind of analysis or if the company will have to innovate even further to achieve its goals. Perhaps the problem is insoluble, event that could be interesting too.

Featured image source: Marco Versch, used under CC License