Abhijit Gupta

Average confidence helps understand how much signal EPA is working with, and when it gets lucky/unlucky. Maybe also include filterable calibration curve on website (stretch goal)

50%+ usage is mobile, make it nicer

Match Breakdown, surrogate/breakdown, event insights, etc.

Simulates per team instead of per match. Leads to teams in the same match having possibly different tiebreakers. For example, 118 should have had no chance of overtaking 2056 in...

Doesn't exclude upcoming matches, see 1922 in 2023.

Add Strength of Schedule/Simulation to DB or API

1

If adding to DB creates too many columns, can expose Python functions that load the data from the DB and run the algorithms.

Maybe add ~10 slots to the database to track arbitrary stats year over year.

Alternatively, EPA with slower/faster update rate for different use cases.