Perpetuals.com, the firm founded by former FTX executive Patrick Gruhn, has launched a new AI-driven trading product designed to turn crowd-sourced strategy signals into profit without requiring users to risk their own capital, according to a report by Bloomberg.
The platform, branded UpsideOnly, allows users to submit simulated trading ideas across markets such as equities, commodities and energy, which are then evaluated by an in-house artificial intelligence system. The strongest signals are executed in live markets using the company’s own capital, with profits shared between the firm and the contributors whose strategies are selected.
The model effectively positions Perpetuals as a hybrid between a hedge fund and a data platform, where retail and professional participants supply predictive inputs rather than direct capital. The firm says its AI system, which has been trained on tens of billions of historical trades, is designed to identify and rank the most consistently profitable strategies emerging from the crowd.
Gruhn, who previously led FTX’s European operations, argues that while markets themselves are inherently difficult to predict, patterns in human trading behaviour can be systematically analysed and monetised. The approach aims to combine human intuition with machine-driven filtering to improve signal quality while leaving execution risk with the firm.
The structure echoes elements of hedge fund research pipelines, where external idea generation is increasingly being systematised and scored using machine learning tools. However, Perpetuals extends this concept by fully internalising trading risk, deploying its own balance sheet rather than client capital.
The company says it is in discussions to raise several hundred million dollars to scale its proprietary trading operations, and Gruhn has also committed significant personal capital to the strategy.
The launch comes against the backdrop of ongoing experimentation across the hedge fund and fintech sectors, where firms are increasingly blending artificial intelligence, crowdsourced data, and proprietary capital to develop new trading models. Despite the enthusiasm, similar approaches in both quantitative trading and AI-driven investing have historically faced challenges in consistently outperforming traditional strategies.
Gruhn has framed the initiative in part as a response to lessons learned during the collapse of FTX, arguing that the new structure avoids client capital exposure while still harnessing collective market intelligence.
