Skip to main content

​
What Trepa does at the base layer

Trepa uses accuracy-weighted pari-mutuel pools for numerical outcomes. Everyone puts their stake into one shared pool and competes based on how close their prediction is to the final outcome. There is no house taking the other side, no order book, and no market maker. Every pool settles internally between participants. Trepa focuses on real-world outcomes like inflation, prices, earnings, or other numeric indicators. You always predict a number, not YES/NO.

​
How it works

1

Join a live pool

You pick a live pool, enter your numerical prediction, and choose how much to stake. You can only submit or modify your prediction while the pool is live.
2

Resolution

Once the outcome is available, Trepa reads the outcome from a verifiable data source. A link to the source is available on the pool page so anyone can verify it.
3

Dispute window (coming soon)

After the outcome is released, there’s a period where any participant can challenge the reported outcome. This triggers a review process before results are finalized. Most pools resolve without disputes.
4

Trepa scores your accuracy

Once the dispute window closes (or all disputes are resolved), the final outcome is confirmed and locked. Trepa then measures the error between each prediction and the outcome to convert it into an accuracy score.
5

Trepa assigns your share

Trepa calculates your payout based on your accuracy score, stake amount, and how early you have made your prediction.

​
How Trepa rewards you

Trepa is designed so that being closer to the outcome can give you much larger rewards through convex accuracy-based payouts, while ROI caps and budget-balance keep the system bounded and self-funding.
  • Convex accuracy curve: as your prediction gets closer to the actual outcome, your accuracy score increases at an exponential (faster) rate.
  • ROI caps: each prediction pool has a capped return on investment that scales with how difficult making a perfect prediction is, so no single participant can eat the entire pool.
  • Time weight: making earlier predictions improves your weight in the pool through a concave time decay function. This also discourages last-second sniping.