Who actually calls it?
Every settled week, every stated distribution, scored with CRPS — a strictly proper scoring rule. The only way to climb is for your distribution to match reality. Wash trades don't move it. Size doesn't move it.
Scoring math is live and settled against real weekly closes; the forecaster population is simulated until testnet accounts land.
How scoring works
CRPS(Continuous Ranked Probability Score) compares your stated distribution's CDF against the settled outcome, bucket by bucket. Lower is better; a perfect point-mass on the winning bucket scores 0. Skillis your improvement over the uniform "no idea" forecast — 0% means you know nothing, 100% would be omniscience.
CRPS is strictly proper: your expected score is optimized only by reporting what you truly believe. That's the anti-farming property — there is no volume grind, no wash-trade loop, no size advantage that improves calibration. If rewards ever flow from this board, they flow to forecasters who were provably right. Stake your own shape →