The DeepMind trio who built a poker AI are now making money for quant hedge funds
TechCrunch
β’Tue, 30 Jun 2026 20:33:48 +0000
π° What Happened
Three former DeepMind researchers have successfully applied reinforcement learning technology originally developed for poker-playing AI to algorithmic stock trading. Their Prague-based company, EquiLibre Technologies, is now valued at $500 million following a Series A funding round led by Creandum β the largest single investment the VC firm has ever made. CEO Martin Schmid noted that the scoring for trading is elegantly simple: how much money the agent makes. In partnership with quant firm Tower Research Capital, EquiLibre's algorithms have been trading billions of dollars in daily volume across the S&P 500 and Nasdaq. The startup claims a perfect record of zero negative months since inception, having first deployed on crypto markets in 2025 before expanding to stock exchanges. EquiLibre's reinforcement learning approach is particularly well-suited to financial markets because it allows self-learning models to optimize for clear reward signals β profit and loss.
π The Backstory
The three founders β Martin Schmid, Matej Moravcik, and another colleague β previously worked at DeepMind (Google's AI research division) where they developed groundbreaking poker AI systems including DeepStack and Player of Games, which achieved superhuman performance at imperfect-information games. These games are conceptually similar to financial markets, where players must make decisions with incomplete information. Quant hedge funds have been using algorithmic trading for decades, but recent advances in reinforcement learning and large-scale neural networks have opened new possibilities. Tower Research Capital, EquiLibre's partner, is one of the oldest quant trading firms, founded in 1998.
π― Why It Matters
EquiLibre's success represents a landmark crossover between cutting-edge AI research and financial markets, demonstrating that reinforcement learning techniques developed in game-playing contexts can generate real economic value. Their $500 million valuation and perfect monthly trading record signal that AI-driven quantitative trading is entering a new phase of sophistication, potentially reshaping how hedge funds operate.
Three former DeepMind researchers have successfully applied reinforcement learning technology originally developed for poker-playing AI to algorithmic stock trading. Their Prague-based company, EquiLibre Technologies, is now valued at $500 million following a Series A funding round led by Creandum β the largest single investment the VC firm has ever made. CEO Martin Schmid noted that the scoring for trading is elegantly simple: how much money the agent makes. In partnership with quant firm Tower Research Capital, EquiLibre's algorithms have been trading billions of dollars in daily volume across the S&P 500 and Nasdaq. The startup claims a perfect record of zero negative months since inception, having first deployed on crypto markets in 2025 before expanding to stock exchanges. EquiLibre's reinforcement learning approach is particularly well-suited to financial markets because it allows self-learning models to optimize for clear reward signals β profit and loss.
The three founders β Martin Schmid, Matej Moravcik, and another colleague β previously worked at DeepMind (Google's AI research division) where they developed groundbreaking poker AI systems including DeepStack and Player of Games, which achieved superhuman performance at imperfect-information games. These games are conceptually similar to financial markets, where players must make decisions with incomplete information. Quant hedge funds have been using algorithmic trading for decades, but recent advances in reinforcement learning and large-scale neural networks have opened new possibilities. Tower Research Capital, EquiLibre's partner, is one of the oldest quant trading firms, founded in 1998.
EquiLibre's success represents a landmark crossover between cutting-edge AI research and financial markets, demonstrating that reinforcement learning techniques developed in game-playing contexts can generate real economic value. Their $500 million valuation and perfect monthly trading record signal that AI-driven quantitative trading is entering a new phase of sophistication, potentially reshaping how hedge funds operate.