NOVA Blueprint Update: The Triple Crown
We’re improving the blueprint’s winner selection mechanism to better reward strong, generalizable performance.
Previously, a challenger could replace the current king by clearing the winning threshold in a single epoch. Under the new **Triple Crown** system, a challenger must clear the threshold across **three consecutive epochs**, each with a **different target**.
The update is intended to:
* Reward improvements that are robust and repeatable
* Encourage stronger generalization across targets
* Increase confidence that a new winner represents a meaningful step forward
* Make successful takeovers more representative of sustained performance
To take the crown, a challenger must now earn it three times against the previous best.
You can review the implementation here: