Perpetuals launches UpsideOnly, an AI-powered ‘risk-free’ market prediction platform
Perpetuals.com Ltd, a Nasdaq-listed fintech group, has introduced UpsideOnly, a market prediction platform that lets users submit forecasts across equities, crypto, commodities and forex without staking personal capital. The company says it will execute trades with its own funds based on a proprietary AI model and distribute profits to users whose predictions help generate successful trades.
Perpetuals describes the service as a human-AI collaboration: participants contribute signals and the platform’s algorithm, BayesShield AI, applies those signals to make trading decisions. According to the company, BayesShield was trained on more than 22 billion retail trades, a dataset the firm says informs real-time signal refinement.
How UpsideOnly works
Users register and make directional predictions on various markets without depositing trading capital. Perpetuals evaluates those signals and, when its AI assigns sufficient probability to a forecast, the firm trades with its own capital. If a trade is profitable, the firm shares some portion of the gains with the users whose signals contributed to the decision. If a trade loses, Perpetuals absorbs the loss and users take no financial hit from the trade itself.
The product offers optional refundable deposits of $1 or more. Perpetuals says those deposits are not used for trading but are held in U.S. Treasury securities with an external fiduciary and are designed to deter bots. Users who place a deposit reportedly receive higher payout levels, reflecting the platform’s view that small deposits improve signal quality.
What the company claims and what to watch
Perpetuals positions UpsideOnly as a corrective to what it characterizes as an unfair retail trading ecosystem. The company argues that letting users contribute insight rather than capital aligns incentives differently from conventional retail broker models, where many individual traders lose money.
Key company claims include the scale of the training data behind BayesShield and the platform’s multi-asset coverage. Perpetuals also notes regulatory connections: an affiliate, PM MTF Ltd., operates under an EU multilateral trading facility regime and the firm cites compliance frameworks such as MiFID II, MiCA, DORA and EMIR for its European operations.
Market and regulatory implications
UpsideOnly sits at the intersection of several fast-evolving fintech trends: AI-driven trading, crowd-sourced forecasting and novel consumer products that blur the line between prediction markets and brokerage services. That mix could draw heightened regulatory attention.
Regulators will likely scrutinize how UpsideOnly is classified under securities, derivatives and gambling laws. The firm itself flags potential risks that the product could be characterized under gaming, sweepstakes or commodity-derivatives regimes, a point that reflects the legal complexity of monetizing user predictions when a firm executes trades on the back end.
Consumer protection authorities may also probe disclosure, the robustness of anti-abuse measures, and whether the model could incentivize risky behaviour from participants who do not face downside. The presence of refundable deposits intended to deter bots may not be sufficient, in regulators’ view, to address market manipulation or wash-trading risks without additional controls and transparency.
Business model and sustainability questions
At its core, the UpsideOnly model transfers trading downside to Perpetuals while offering users upside participation. That places a premium on three elements: the predictive edge of BayesShield, effective risk management at the firm level, and the durability of the signal sourcing mechanism.
Model performance and data limits. The company cites a very large retail trade dataset, but past performance and backtests do not guarantee future returns. Machine learning models trained on historical retail behaviour can be vulnerable to regime shifts, data biases and overfitting. The platform’s long-term profitability depends on whether the AI can identify persistent edges after costs and the profit split to users.
Capital and risk appetite. Perpetuals must bankroll losing streaks and market shocks, which could strain capital if the platform scales quickly or encounters rare adverse events. The firm’s Nasdaq listing gives it public-market access to capital, but sustained losses or drawdowns could prompt investor scrutiny.
Gaming the system. Prediction marketplaces and incentive systems attract strategic actors. Participants who try to exploit payout mechanics or coordinate signals could degrade signal quality and create losses. The refundable deposit is a modest deterrent and may reduce automated accounts, but it is not a comprehensive anti-abuse solution.
Industry context
UpsideOnly is part of a broader wave of products attempting to democratize alpha and repackage trading as an experience anchored by gamified inputs or social signals. The difference here is the explicit claim that retail users are shielded from trading losses because the firm executes trades with its own balance sheet.
That arrangement recasts users as signal providers rather than capital providers. Economically, that is similar to some prediction markets and algorithmic investment strategies that pay contributors for valuable information. Whether the economics scale as user volumes rise will be an important test of the concept.
Bottom line
Perpetuals’ UpsideOnly brings together crowd intelligence and an AI model trained on large retail datasets to offer participants exposure to trading gains without direct downside. The architecture raises meaningful questions about regulatory treatment, model risk, and capital sustainability. Market participants and regulators will be watching closely to see whether Perpetuals can convert user signals into repeatable returns while managing the operational, legal and financial risks inherent in underwriting trading losses.
Bloomberg has also reported on the launch, and Perpetuals has made its executive team available for interviews. The company cautions that “risk-free” refers specifically to user trading losses and does not eliminate operational or other platform risks.
