Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.trepa.io/llms.txt

Use this file to discover all available pages before exploring further.

Sometimes a round has very few winners or unusually spread-out errors. Without a ceiling, almost the entire prize pool could concentrate on one or two forecasts. That would define everyone else’s experience of the game off rare spikes and skew incentives. The cap smooths those edge cases so rounds stay fair. After accuracy weight splits the prize pool among winners, each winner’s profit that round is capped at 100× their entry fee. Gains are scaled together (water-filling below) so the full pool still pays out—no one exceeds their cap, and nothing is skimmed from the player prize pool into the house when the cap binds.
If the uncapped math would pay you more than that ceiling, the extra does not go to the house. It is rerouted to the other winners in the same round (again by accuracy weight).

​
Hypothetical: where the extra goes

Imagine the prize pool for winners’ profits this round is $220. Before any cap, suppose accuracy-weighted shares would have paid Winner A $180 profit and Winners B–E $40 profit combined ($180 + $40 = $220). With the $100 profit cap on a $1 entry, Winner A receives $100 profit ($101 including entry back), not $180. The $80 that would have gone over the cap is reallocated among B–E by weight, so together they receive $120 profit instead of $40. Total profit paid is still $220; platform take is unchanged. Only the split among winners moved. The exact ceiling is a tuning knob: we set it so bounded upside still leaves room for big rounds to feel meaningful, and we may adjust it if competition and feedback suggest we should.

​
Formulas (for verification)

  • Dividends (prize pool for winners) = losers’ entry fees minus platform take.
  • Gain cap for winner ii: capi=entry feei×100\text{cap}_i = \text{entry fee}_i \times 100.
  • Gain: gaini=min⁡(α×ai, capi)\text{gain}_i = \min(\alpha \times a_i,\, \text{cap}_i), where aia_i is your accuracy weight and α\alpha is a single scale factor chosen so that the sum of all winners’ gains equals the dividend pool while no one exceeds their cap (water-filling).
  • Payout: payouti=entry feei+gaini\text{payout}_i = \text{entry fee}_i + \text{gain}_i for winners; losers receive 0.
Last modified on May 4, 2026