DeepMind poker alumni apply reinforcement learning to trading
former deepmind researchers raised a series a for equilibre technologies, applying reinforcement learning agents to stock trading with a reported zero-loss track record.
equilibre technologies, founded by former deepmind researchers behind the deepstack poker ai, secured a major series a investment led by creandum, pushing its valuation to $500 million. the firm utilizes reinforcement learning—the same technique used to master imperfect information games like poker—to power trading agents deployed with tower research capital. these agents currently trade billions in daily volume on the s&p 500 and nasdaq, with the startup claiming a perfect record of positive monthly returns. by operating in prague with a team of 25, the firm leverages a quantifiable reward signal—profit—to train models without the immense data labeling costs of large language models. this development signals a trend of ai talent moving from pure research into high-value verticals like finance, where reinforcement learning can be immediately monetized.