Retail traders have long struggled in leveraged crypto markets, and nowhere is this more visible than in Bitcoin perpetual futures. During volatile market phases, a large share of non-professional traders consistently lose money, often due to emotional decision-making, over-leverage, or poor timing. A new pilot project by Perpetuals.com suggests that artificial intelligence may be able to identify — and even prevent — many of these losses before they occur.
Perpetuals.com Ltd. (Nasdaq: PDC), a regulated fintech company combining blockchain infrastructure with advanced AI, has announced the successful completion of its BayesShield™ AI pilot program. According to the company, the system was able to detect 92 percent of loss-making retail trades in the BTC-USD perpetual futures market during backtesting.
The pilot analyzed real trading decisions made by retail traders over the past year. By applying proprietary machine learning models to twelve months of historical Bitcoin perpetual futures data, BayesShield™ successfully flagged the vast majority of trades that ultimately resulted in losses. The results suggest that retail trading losses in leveraged crypto markets are not random, but follow identifiable and repeatable patterns.
Patrick Gruhn, CEO and founder of Perpetuals.com, says the findings confirm years of research into retail trading behavior. In his view, most retail losses in leveraged derivatives markets are both predictable and avoidable. BayesShield™, he argues, represents a fundamental shift in how trading platforms could protect users while still maintaining fair and efficient markets.
The timing of the pilot is notable. The crypto market has recently gone through sharp corrections, testing trader discipline and risk management across the board. In such conditions, predictive systems that can identify high-risk trades before execution could offer meaningful protection. By blocking or warning against trades with a high probability of failure, BayesShield™ aims to reduce retail losses while strengthening overall market stability.
Beyond individual trades, the system evaluates each trader’s profit-and-loss profile and combines it with broader market sentiment signals. This allows BayesShield™ to calculate win-loss probabilities in real time. The technology is designed not only for retail protection, but also for use by market makers, derivatives platforms, and structured product issuers seeking to optimize hedging strategies and manage risk more effectively.
Historically, retail derivatives markets have operated on a model where most participants lose over time. Gruhn believes AI can change that dynamic. By predicting which trades are likely to fail before they are executed, trading environments could become more ethical, sustainable, and resilient — an increasingly important factor as regulators worldwide scrutinize retail crypto derivatives.
Following the pilot’s success, Perpetuals.com has moved into full-scale development of the BayesShield™ production platform. The system is being trained on one of the largest retail trading datasets assembled to date, comprising billions of individual trades across crypto spot markets, perpetual futures, options, leveraged products, and even traditional equity markets. Training is conducted on NVIDIA-powered infrastructure to handle the scale and complexity of the data.
The announcement also comes shortly after Perpetuals.com completed its technical integration following a business combination with EarlyWorks Co., Ltd. The merger unified technology stacks, data systems, and operational infrastructure under a single platform. The company’s redesigned website now serves as a central hub for its exchange technology, custody services, cross-chain infrastructure, payment solutions, and tokenization offerings.
As regulators increase pressure on high-loss retail derivatives products, BayesShield™ positions Perpetuals.com as a company attempting to address both systemic inefficiencies and ethical concerns in the sector. The firm plans to deploy the system within its regulated trading infrastructure, including a third-party operated, CySEC-authorized MiFID II licensed multilateral trading facility, with broader commercial rollout expected in the second half of the year.
Whether AI-driven trade prevention becomes a new industry standard remains to be seen. But if systems like BayesShield™ deliver on their promise, the future of Bitcoin derivatives trading may be less about who loses fastest — and more about who trades smarter.