Superteam Demo Day: Trepa (Jong-Chan Chung)

What if prediction markets rewarded you for being accurate rather than just being right? A new Solana-based platform called Trepa is challenging the fundamental mechanics of how prediction markets work, introducing a system where getting closer to the actual outcome means exponentially higher returns.

Summary

Trepa, presented by co-founder Jong-Chan Chung at Superteam Demo Day during Breakpoint 2025, represents a paradigm shift in how prediction markets function. Instead of traditional binary yes/no bets, Trepa allows users to predict specific numerical outcomes across various domains including prices, earnings, macroeconomic indicators, election margins, and sports statistics.

The platform's core innovation lies in its reward mechanism: users who make predictions closest to the actual outcome receive exponentially higher returns. This addresses a fundamental flaw in traditional prediction markets where someone can be punished for being highly accurate but directionally wrong, while another participant wins despite being further from the truth simply because they guessed the correct side of a binary question.

Launched on Solana just four days before the Breakpoint presentation, Trepa has already begun generating what the team describes as "valuable and high-resolution data" that they believe complements and even feeds upstream to existing prediction market systems. The platform features a simple slide-to-predict interface where users stake their conviction on specific numbers and earn rewards if their predictions fall within the top 50% closest to the actual outcome.

The founding team consists of two German co-founders who met in the prediction market space, bringing technical expertise in building trading alerts for prediction markets and experience predicting the effects of monetary policies. The project has secured backing from notable investors in the space, signaling confidence in this novel approach to decentralized predictions.

Key Points:

The Problem with Binary Prediction Markets

Traditional prediction markets operate on a simple yes/no framework that fundamentally misrepresents how the real world works. Jong-Chan Chung illustrated this problem with a compelling example: imagine a prediction market asking what the highest temperature will be next year. If the actual outcome is exactly 25 degrees, someone who predicted 24 loses their entire stake, while someone who predicted 31 wins—despite being much further from the truth.

This creates a perverse incentive structure where precision is punished when directionally incorrect. A user could demonstrate remarkable analytical skill by being just one unit away from the correct answer, yet receive nothing for their accurate insight. Meanwhile, someone with far less precision but the "correct" directional guess walks away with the winnings. Trepa directly addresses this fundamental design flaw by rewarding closeness to outcomes rather than binary correctness.

Exponential Rewards Based on Proximity

Trepa's reward system is built on a simple but powerful principle: the closer you are to the actual outcome, the higher your return on investment—exponentially so. This creates meaningful incentives for users to apply genuine analytical rigor to their predictions rather than simply gambling on binary outcomes.

The exponential nature of the rewards means that extremely precise predictions are disproportionately valuable, encouraging serious forecasters to participate while still providing opportunities for those who make reasonably close predictions. Users whose predictions fall within the top 50% closest to the actual outcome receive payouts, with the distribution heavily favoring the most accurate predictors.

Real-World Experiment at Breakpoint

To demonstrate the platform's concept in action, Trepa conducted an engaging experiment during Breakpoint 2025. They asked 50 conference attendees to predict the exact number of M&Ms in a jar, creating a tangible, verifiable prediction market scenario.

The results revealed fascinating insights about crowd wisdom and prediction accuracy. The actual count was 979 M&Ms, and the closest prediction came in at 975—remarkably close given the difficulty of the task. Perhaps more interesting to the team was that the distribution of all estimates formed a bell curve centered around the actual outcome, demonstrating the potential of aggregated numerical predictions to generate meaningful, high-resolution data about future events.

Data Value and Market Positioning

Trepa positions itself as generating data that is both complementary to and upstream of existing prediction market systems. This suggests a strategic vision where Trepa's high-resolution numerical data could serve as a foundation for more sophisticated prediction ecosystems.

Traditional binary prediction markets produce relatively coarse data—essentially probabilities for discrete outcomes. In contrast, Trepa's approach generates distribution curves and precision metrics that could provide richer insights into market sentiment and collective forecasting accuracy. This data positioning could make Trepa valuable not just as a consumer-facing product but as infrastructure for the broader prediction market ecosystem.

Simple User Experience on Solana

Despite the sophisticated mechanics underlying the platform, Trepa emphasizes simplicity in its user interface. Users interact with the platform through a straightforward slide mechanism to select their predicted number, then stake tokens representing their conviction level in that prediction.

Launching on Solana provides the transaction speed and low costs necessary for a prediction market that deals with potentially many fine-grained numerical predictions rather than coarse binary bets. The four-day-old deployment ahead of Breakpoint suggests the team moved quickly to have a live product for the demo day presentation.

Facts + Figures

• Trepa launched on Solana just four days before their Breakpoint 2025 presentation

• The platform rewards users whose predictions are in the top 50% closest to the actual outcome

• Returns scale exponentially based on prediction proximity—closer predictions earn disproportionately higher rewards

• During a Breakpoint experiment, the actual M&M count was 979, with the best prediction being 975

• 50 Breakpoint attendees participated in the M&M prediction experiment

• The distribution of predictions formed a bell curve around the actual outcome

• The platform accepts predictions across multiple categories: prices, earnings, macroeconomic indicators, election margins, and sports statistics

• Trepa was founded by two German co-founders who met in the prediction market space

• One co-founder has built trading alerts for prediction markets; the other has experience predicting monetary policy effects

• The project has secured backing from investors in the prediction market and crypto space

Top Quotes

• "Get closer, win bigger."

• "Binary bets kill precision."

• "Essentially, you are punished for being accurate but directionally wrong."

• "The closer you are to the real outcome, the higher your ROI. Exponentially."

• "Simply slide to your number, stake your conviction and win cash in cloud if you're in the top 50% of close predictions."

• "The distribution of estimates took the shape of a beautiful bell curve around the actual outcome."

• "Over time, we believe that the data that we are generating is valuable and high-resolution as well as complementary and upstream to prediction market systems."

Questions Answered

How does Trepa differ from traditional prediction markets?

Traditional prediction markets force users to make binary yes/no decisions, which often misrepresents the nuanced nature of real-world outcomes. Trepa replaces this binary model with numerical predictions where users submit specific numbers rather than choosing sides. The reward system then distributes winnings based on how close each prediction was to the actual outcome, with closer predictions earning exponentially higher returns. This means accuracy is always rewarded, even if a prediction was on the "wrong side" of a median value.

What types of predictions can you make on Trepa?

Trepa supports a wide range of numerical prediction categories spanning finance, politics, and sports. Users can make predictions about prices, corporate earnings, macroeconomic indicators, election margins, and sports statistics. Any scenario where the outcome can be expressed as a specific number is potentially suitable for Trepa's prediction model. This versatility allows the platform to serve diverse user interests while generating varied data across multiple domains.

How do rewards work on the platform?

Rewards on Trepa are calculated based on proximity to the actual outcome, with an exponential scaling mechanism that heavily favors the most accurate predictions. If your prediction lands within the top 50% closest to the actual result, you receive a payout. The exponential nature means someone who predicts within one unit of the outcome receives dramatically more than someone who was ten units away, even if both technically "won." This creates strong incentives for careful analysis rather than random guessing.

Why did Trepa choose to build on Solana?

While the presentation didn't elaborate extensively on the Solana choice, prediction markets require frequent transactions, fast finality, and low fees—all areas where Solana excels. A numerical prediction market generates more granular data than binary markets, potentially requiring more on-chain activity. Solana's high throughput and low transaction costs make it well-suited for this use case, allowing users to participate without prohibitive gas fees eating into potential winnings.

What makes Trepa's data valuable?

The data generated by Trepa offers higher resolution than traditional prediction markets because it captures full distributions of numerical predictions rather than simple probability estimates for binary outcomes. This distribution data, as demonstrated by the bell curve that emerged from the M&M experiment, provides richer insights into collective forecasting behavior. The team positions this data as complementary to and even upstream of existing prediction market systems, suggesting potential applications in market research, forecasting models, and decision-making tools.

Who is behind Trepa?

Trepa was founded by two German co-founders who met while working in the prediction market space. One founder has a background in full-stack engineering and previously built trading alerts for prediction markets, while the other brings experience in predicting the effects of monetary policies. This combination of technical capability and domain expertise positions them well to build a sophisticated prediction platform. The project has also attracted backing from notable investors in the crypto and prediction market space.