Skip to main content
This guide walks you through the process of creating your own Crunch by forking and customizing the starter kit.

​
Fork the Starter Kit

Start by forking the Coordinator Node Starter Kit repository: 
git clone https://github.com/crunchdao/coordinator-node-starter my-crunch
cd my-crunch

​
Define Your Prediction Task

Your prediction task defines what participants need to predict. This includes:
  • Input data — What data will models receive?
  • Output format — What should predictions look like?
  • Timing — How often are predictions made?

​
Define Your Scoring Function

The scoring function determines how predictions are evaluated and ranked. Consider:
  • Evaluation metric — How do you measure prediction quality?
  • Aggregation — How do individual scores combine into rankings?
  • Time windows — What time periods matter for scoring?

Next: Deploy to Testnet

Once your Crunch is working locally, deploy it to testnet for broader testing.