A 2022 report described three former DeepMind researchers building an AI system intended to trade stocks and cryptocurrencies. It did not show that the system could reliably spot coins before they rose, publish audited returns, or become an investment product ordinary readers could use. “Invest in crypto before they rise” was an ambition, not a demonstrated capability.
Who was behind the EquiLibre project?
Tech Times reported on April 5, 2022, that Martin Schmid, Rudolf Kadlec, and Matej Moravcik had left DeepMind in January and formed EquiLibre Technologies in Prague. The report said the three had also worked at IBM and had relocated from Edmonton, Canada. It did not provide full employment histories, individual roles, or ownership details. Tech Times’ 2022 report is the available source for these details.
Their earlier work included DeepStack, a poker-playing AI. The report described DeepStack as the first AI to defeat professional players in heads-up no-limit poker in 2017. That history helps explain the founders’ interest in sequential decision-making, but it is not evidence that their later system could forecast financial markets.
What were they trying to build?
EquiLibre was described as a startup applying machine-learning techniques to financial markets, with both stocks and cryptocurrencies under consideration. Schmid said the team was training a system to make buy-and-sell decisions for profit. The report framed the work as development, not a completed service or proven trading strategy.
#1 Best Overall
The approach was associated with reinforcement learning: a model takes actions, receives feedback from its environment, and adjusts toward a chosen reward. In principle, a trading system could choose whether to buy, sell, hold, set position sizes, or allocate a portfolio. Its reward function would need to account not just for nominal profit but also for losses, trading costs, excessive turnover, leverage, drawdowns, and concentration. The original report did not disclose EquiLibre’s architecture, training data, assets, trading horizon, reward function, execution venues, or risk controls, so its exact design cannot be assessed from the public account.
Why poker experience does not prove trading skill
Poker and trading both involve uncertainty and decisions made over time, which makes the analogy appealing. But poker has a defined game, rules, and outcomes; financial markets change as participants, liquidity, regulations, and conditions change. A system that performs well in poker has not thereby shown it can predict asset prices or earn returns after costs.
Rank #2
- Comes with secure packaging
- Easy to read text
- It can be a gift option
In markets, a backtest can look strong because a model has learned noise or accidentally used information unavailable at the time of a trade. A pattern can also stop working once conditions change or other participants exploit it. Even a correct directional signal may fail to make money after spreads, fees, slippage, market impact, funding costs, or unfilled orders.
What was claimed, and what was actually established?
The 2022 report attributed several plans and views to Schmid: the team wanted to adapt ideas from poker AI to trading, believed it could improve on existing algorithms, and hoped eventually to create a fund or sell the technology to a financial institution or another investor. It also attributed to him a claim that EquiLibre had raised the largest-ever Czech seed round, without disclosing an amount. These are reported statements and ambitions, not verified performance or proof of a completed financing outcome.
The report did not provide an audited return series, benchmark comparison, Sharpe ratio, maximum drawdown, live account record, or independently reproduced experiment. It also did not establish that a public app, fund, or institutional trading product launched. As of August 18, 2026, the available reporting does not verify EquiLibre’s later status or show that ordinary readers can invest through it.
- A private startup is not the same as a publicly investable company.
- A model under development is not a live trading record.
- A backtest is not proof of returns in actual markets.
- An AI-generated signal is not a guarantee that an asset will rise.
Why an AI trading system faces practical hurdles
Changing markets and misleading backtests
Market behavior is nonstationary: relationships that appeared useful in one period may disappear as liquidity, regulation, or participant behavior changes. Reliable evaluation therefore needs data held back from training, walk-forward testing, and controls against data leakage. Results should be compared with a defined benchmark over stated periods and risk levels.
Rank #4
Costs, liquidity, and execution
Crypto trades around the clock across fragmented venues. Thin liquidity, price gaps, exchange outages, API failures, liquidation cascades, and partial fills can make simulated executions unlike real ones. Custody and counterparty risks, stablecoin problems, hacks, token suspensions, and delistings can also affect outcomes independently of a model’s forecast.
Reward design and risk limits
A system rewarded only for profit may take concentrated positions, trade excessively, or use leverage in ways that increase the chance of catastrophic loss. A credible design would need explicit constraints for position size, leverage, turnover, drawdown, and liquidity—not just a profit target. The report did not describe whether EquiLibre had such safeguards.
Recommended Free Tools
Best Value
Operations and regulation
Research software, selling signals, automated execution, managing a fund, and giving personalized advice are different activities with different operational and regulatory considerations. Requirements also depend on jurisdiction and product. Schmid’s reported lack of concern about regulation was not a legal analysis, and the report did not identify the jurisdictions or business model in which the system would operate.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Could readers invest in EquiLibre or use its system?
The reporting did not say that readers could buy EquiLibre shares, subscribe to its strategy, or invest in a public fund. Its commercialization ideas were future possibilities. It would be misleading to treat the article as an offer or recommendation to invest.
Anyone evaluating an AI trading claim should look for net results after costs, out-of-sample and live performance, a clear benchmark, drawdowns, the amount of capital traded, and independent verification. Without those details, claims about picking future winners remain unproven.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →
