The transition from smoke-filled poker rooms to the sterile server halls of Wall Street is no longer just a theory. EquiLibre Technologies, a startup founded by Google DeepMind veterans, is proving that the ability to bluff and outmaneuver opponents is exactly what traditional quantitative trading was missing. The team that taught AI to crush professional poker players has now adapted reinforcement learning architectures for the stock market. The result: a $500 million valuation following a Series A round led by Creandum. According to Creandum VP Cameron Sellers, this is the largest single investment in the firm's history—a massive bet on a new technological paradigm.
Game Mechanics at the Service of Capital
EquiLibre’s technical foundation rests on the parallels between poker and market dynamics—specifically, decision-making under conditions of imperfect information. As CEO Martin Schmid explains, reinforcement learning is a perfect fit for trading because the reward signal—net profit—is absolute and unambiguous.
In partnership with Tower Research Capital, EquiLibre’s agents are already processing billions in volume across the S&P 500 and Nasdaq. Since launching in crypto markets in 2025, the company claims a phenomenal track record: not a single loss-making month.
Aggressive Scaling
While EquiLibre carefully maintains the persona of a scientific laboratory, its commercial trajectory is shark-like in its aggression. The founding team, including CTO Rudolf Kadlec and Chief Science Officer Matej Moravčík, is scaling operations with startling speed. This serves as a direct warning to asset managers and investors: if your strategies still rely on static historical data analysis rather than real-time adaptive learning, you risk sitting at a table where DeepMind experts are about to take your entire stack.
Watch your portfolio’s performance drift closely. The arrival of agents trained on game theory in high-liquidity equity tiers isn't just another software update—it is a direct threat to your alpha.
In a market where machines are learning to win under uncertainty, legacy statistical methods are quickly becoming easy prey.