The Prediction Market That Doesn’t Take Your Money
In January, Polymarket processed roughly four million euros in wagers on the Portuguese presidential election in the hours before results were announced. Within forty-eight hours, the country’s gambling regulator ordered the platform to wind down its operations in Portugal. Hungary’s regulator blocked the domain entirely the same week. By the end of the month, Polymarket had been progressively banned, restricted, or threatened with enforcement action across more than a dozen European jurisdictions, including Germany, France, Belgium, the Netherlands, and Switzerland.
The pattern has become familiar. A prediction-market platform reaches enough volume to attract attention. National regulators classify it as unlicensed gambling. Access is blocked, sometimes with fines threatened. The platform geofences and moves on. Regulators in Paris, Berlin, Lisbon, and Amsterdam have all reached versions of the same conclusion: contracts that pay users based on the outcome of real-world events look like gambling, regardless of how the platform describes itself.
Against that backdrop, a New York-based startup called SafeBets has launched what may be the most architecturally unusual entrant the prediction-markets sector has seen. SafeBets users predict the future pricing of widely traded assets—Bitcoin, gold, oil, shares of major corporations—but they do so without putting any money at risk. They cannot lose. The only thing they can earn is a cryptocurrency called Unicoin, awarded to top performers at the platform’s discretion.
That structural difference—no user wagers, no contractual payouts, no financial exposure—may turn out to matter more than its founders are publicly emphasizing. In jurisdiction after jurisdiction across Europe and Asia, regulators have banned competitors precisely on grounds that, by SafeBets’s own account, do not apply to its model. If that argument holds, SafeBets stands to inherit by default the markets its rivals are being expelled from.
How It Works
The mechanics are straightforward. Users sign up at SafeBets.world, complete identity verification, and receive 100 unicoins as a starting balance. They use those unicoins to make predictions on the future pricing of most traded items: Bitcoin, gold, oil, shares of major corporations. If a prediction is right, the user accumulates more unicoins. If it is wrong, the user simply loses one unicoin from their starting allotment of free unicoins.
Critically, no user ever deposits money. The unicoins were given to them. They cannot be purchased on the platform. Users do not buy in, do not stake capital, and have no contractual right to any specific outcome.
The unicoins themselves are units of Unicoin, the network token underlying the SafeBets platform. Konanykhin describes Unicoin as the Smart Coin for Smart People—a cryptocurrency designed to be mined not through brute-force computation, in the manner of Bitcoin, but through what the company calls proof-of-intelligence. In practice, that means unicoins are awarded to users who consistently make accurate predictions, with the algorithm functioning as a filter that surfaces and rewards genuine forecasting skill. The economic logic is meant to be self-reinforcing: better predictors earn more unicoins, the platform’s aggregated signal improves, and the value of the underlying network token grows in step with the quality of the community using it.
Where does the money behind the prizes come from? SafeBets aggregates the predictions of its highest-ranked users—identified through what the company calls a Filtration Pyramid that ranks predictors by historical accuracy—and uses the resulting collective signal to inform its own trading in real financial markets. When the platform’s trading is profitable, half of the profits are paid out to top predictors as unicoin awards. The unicoins are convertible to cash through the broader Unicoin market.
Alex Konanykhin, SafeBets’s CEO, describes the model in plain terms. The platform is built to harness what economists have long called the wisdom of crowds, with the additional refinement that not all crowds are equal. Most platform users will be wrong most of the time. A small minority will be reliably accurate. The company’s algorithm is designed to find the second group and listen to them.
“Our users risk nothing. The platform takes the risk, makes the profit, and shares it with the people whose insights generated the trade.”
That last sentence is also the company’s regulatory argument.
Why Regulators May Not Have Grounds to Ban It
Polymarket and Kalshi, the two largest prediction-market operators, have spent the past two years arguing in courtrooms and regulatory hearings that their products are financial markets rather than gambling venues. The argument has succeeded at the U.S. federal level—the Commodity Futures Trading Commission approved Polymarket’s return to the United States in late 2025, and Kalshi operates as a CFTC-regulated Designated Contract Market—but it has fared poorly almost everywhere else. State gaming regulators in Nevada, Tennessee, and Massachusetts have taken enforcement action against various operators. European regulators have been more uniform: France’s ANJ, Germany’s GGL, the Netherlands’ KSA, Portugal’s SRIJ, and Switzerland’s Gespa have all classified prediction-market activity as unlicensed gambling.
In every one of those rulings, the operative legal element has been the same: users wager money on uncertain outcomes. That is what gambling law in most jurisdictions actually prohibits. Strip out the wager, and the legal hook regulators have been using disappears.
This is the structural difference SafeBets is built around. Where Polymarket says “this is a financial exchange, not gambling,” SafeBets says, in effect, “this is neither, because there is no enforceable contract and no user money at risk.” Users place no wagers, sign no financial contracts, and have no right to any specific payout. Prizes are discretionary. The company’s own trading activity—conducted on its own account in real markets—is regulated separately under whatever rules apply to proprietary trading wherever it occurs.
The practical consequence is significant. In each of the dozen-plus European jurisdictions where Polymarket has been blocked, the regulator’s stated rationale was unlicensed gambling. Whether SafeBets’s no-wager architecture will survive scrutiny in those same jurisdictions is an open question—no European regulator has yet ruled on a discretionary-prize prediction platform with no user wagers, because no significant platform has previously offered one. But the legal grounds that have been used to expel SafeBets’s competitors do not, on their face, reach a model where users risk nothing.
That creates a genuinely unusual competitive position. As Polymarket and Kalshi face progressive exclusion from European, Asian, and Latin American markets, SafeBets is operating in those same markets without having attracted comparable regulatory action. If the architectural argument holds—and it has not yet been tested adversarially—SafeBets may find itself the only meaningful prediction-market option available to users in jurisdictions where the established players have been ordered out. Winning by default is not the same as winning, but in a sector where customer acquisition costs and regulatory friction dominate the economics, the difference can be hard to see from the outside.
Wulf Hambach, a German lawyer specializing in gambling and financial regulation, has argued that there may be regulatory space for prediction markets in Europe “if the product is framed differently—in the direction of financial options trading, for example under the supervision of BaFin.” SafeBets’s framing goes a step further, attempting to position the user-side product as not requiring financial instruments authorization at all. It is an aggressive posture, but a coherent one, and it is being tested in a moment when regulators have already established what they will not allow.
Who Pays for the Prizes
The economics of SafeBets depend on the platform’s trading desk being good enough at trading to generate profits sufficient to fund the prize pool. That is not a trivial assumption. Hedge funds with large research budgets, sophisticated quantitative models, and decades of accumulated expertise often fail to outperform passive benchmarks. SafeBets is betting that aggregating signal from thousands of retail predictors, filtered through an accuracy-ranking algorithm, can produce trading edge that those funds lack.
The intellectual lineage here is real. Robin Hanson, the economist who popularized the term prediction market, has been arguing since the 1990s that aggregated forecasts can outperform expert judgment in specific structured contexts. The Iowa Electronic Markets correctly predicted U.S. presidential election outcomes more accurately than polling averages for years. Corporate prediction markets at firms including Hewlett-Packard, Google, and Microsoft have shown internal forecasting accuracy gains. The empirical case for collective intelligence is reasonably strong.
Whether it translates to consistent alpha in liquid public markets, where every available signal is already being mined by well-capitalized professionals, is a different question. SafeBets’s pitch is essentially that retail predictors collectively notice things that professional traders miss—sentiment shifts, narrative changes, on-the-ground observations—and that the platform’s algorithm can extract usable signal from that noise.
If that proves true, the prize pool funds itself, and the proof-of-intelligence framing becomes more than marketing language: it describes a working system in which forecasting skill is the scarce input and unicoins are the economic output. If it does not, the discretionary nature of the awards becomes operationally important: the platform can simply pay out less in lean periods. That flexibility is part of what makes the legal architecture work, but it also raises a question for users about how reliable the rewards will be over time.
The Wider Context
SafeBets is part of a broader portfolio operated by TransparentBusiness Inc., a New York company formerly known as Unicoin Inc., which is also planning a public token offering for September. The parent company has a publicly disclosed dispute with the U.S. Securities and Exchange Commission related to its earlier Unicoin offerings, which Konanykhin has characterized as regulatory overreach and which remains unresolved at the time of writing. SafeBets itself has not been the subject of SEC action.
The platform is functional and accepting users today. Konanykhin’s stated ambition is large—he has publicly suggested that the structure could, over time, support a Unicoin community larger than any other cryptocurrency, including Bitcoin. Whether that ambition is realized depends on factors well beyond the architectural question, including user acquisition, the consistency of trading performance, and how regulators in major jurisdictions ultimately characterize the model.
What is already true is that SafeBets has put a genuinely different proposition into the prediction-markets sector at a moment when the dominant players are absorbing serial regulatory defeats almost everywhere outside the United States. The architectural argument has not yet been tested adversarially, and a regulator somewhere will eventually try. But the platform launches into a landscape in which the legal grounds used to expel its competitors do not, on their face, apply to it—and into a set of jurisdictions in which those competitors are no longer available to users at all. Whether being structurally different turns out to be enough is the question the next eighteen months will answer.
