Torch Litepaper
WebsiteX (Twitter)
  • Abstract
  • Introduction
    • Problem statement
  • Solution limitations
  • Vision & Objectives
  • How it works
  • Continuous market
  • Forecasting model
  • Probability map
  • Public goods
  • Betting
  • No exits
  • Prediction resolution
  • Hitting the range
  • Payout system
  • Payout formula
  • Reserve management
  • System mechanics
    • Key parameters
    • Lead time quality
    • Local confidence
    • Boldness quality
    • Sharpness quality
    • Bonus share
Powered by GitBook
On this page

Reserve management

To keep the system adaptive and rewarding high-quality predictions while preventing unsustainable reward inflation or reserve depletion, the following strategy is used:

  • At launch, each token sets a Target Level of reserve funds, aimed at maintaining payout models sustainable. It will be set to $1M at project launch for blue chip tokens.

  • As betting activity grows, the reserve funds are accumulated until reaching its Target Level from lost bets. Anything above is subject to payout as a reserve bonus. A healthy balance is when Reserve Bonus is <20% of the total payout on average.

  • Torch continuously monitors historical betting data, tracking Prediction Quality scores and actual outcomes, and reserve funds impact. These statistics inform dynamic tuning of Lead Time, Boldness, and Sharpness configs and weights, as well as Scaling Factor, to maintain a stable Target Level of the reserve.

PreviousPayout formulaNextKey parameters

Last updated 1 month ago