Applying AI To The Stock Market
Written by Sue Gee   
Wednesday, 06 April 2022

Three members of the team that built DeepStack, the first AI system to beat humans at heads-up, no-limit poker, have left DeepMind to form a new startup to apply AI techniques to stock market trading.

If you've read The Fear Index, or seen its TV adaptation, you may well be disturbed by the idea of letting an AI loose to take decisions about algorithmic trading. No good can surely come of it and not only human wealth but even human lives will be put in jeopardy by such a more. However applying reinforcement learning and game theory to "build the next generation of algorithmic trading" is what the newly formed EquiLibre Technologies not only plans to do but appears to be already doing.

EquiLibre's co-founders Martin Schmid, Rudolf Kadlec, and Matej Moravcik left their posts at DeepMind's Edmonton, Canada, facility in January and relocated to Prague, Prior to DeepMind they had worked together at IBM and Schmid and Moravcik had  been part of a larger team that developed DeepStack, the first AI system to beat humans at heads-up, no-limit Texas hold'em poker. This feat, documented in a paper published in Science was achieved in 2016 after playing over 44,000 hands of Poker against 33 professional human players from 17 countries, beating all but one of them by a statistically-significant margin. Over all the games played, DeepStack, an algorithm  outperformed players by over four standard deviations from zero.

Michael Bowling, one of the three members of EquiLiibre's Advisroy Board, who is currently Research Team Lead at the Edmonton DeepMind office and professor at the University of Alberta led the Computer Poker Research Group at Alberta which developed DeepStack. Another board member, Rich Sutton is Professor of Computer Science at Alberta and a Distinguished  Research Scientist at DeepMind while the third, Michal Pechoucek is a professor at the Czech Technical University in Prague, which was also part of the DeepStack effort.

The similarities between playing Poker and playing the stock market must start with the fact that both are characterized by imperfect information. They also share other traits - risk and subterfuge perhaps being the most obvious.

According to the paper DeepStack:

combines recursive reasoning to handle information asymmetry, decomposition to focus computation on the relevant decision, and a form of intuition that is automatically learned from self-play using deep learning.

Speaking to CNBC, Schmidt siad:

"Our idea is that rather than playing poker, our algorithms will play algorithmic trading,”

The intention is to use the reinforcement learning techniques developed in DeepStack to train an AI system to buy and sell shares and make a profit.

Schmid also told CNBC that not concerned about regulators clamping down on the technology as other companies are already doing similar things citing AI algorithmic stock picking products Candlestick and Yuyostox as the technologies EquiLibre will be competing with. Venture capitalists have been keen to back EquiLibre Technologies and Schmid told CNBC that it has raised the largest-ever seed round in the Czech Republic, while not divulging the exact figure.

In the long run, EquiLibre Technologies hopes to either use the AI it develops to underpin a new hedge fund or sell it to a large institutional bank or another investor. Having recently read the  thriller by Robert Harris about a hedge fund using a novel AI-based algorithm, all I can advise is to build in plenty of checks and balances.

smark

More Information

EquiLibre Technologies

DeepStack: Expert-level artificial intelligence in heads-up no-limit poker by

Three former DeepMinders are developing A.I. to pick stocks and crypto

Related Articles

DeepMind Takes On The Rain

DeepMind Solves Quantum Chemistry 

Why AlphaGo Changes Everything

AlphaFold Reads The DNA

AlphaFold Solves Fundamental Biology Problem

AlphaFold DeepMind's Protein Structure Breakthrough

<ASIN:B005EM8O8O>

<ASIN:0307957934>

 

Last Updated ( Wednesday, 06 April 2022 )