The Forecast Exchange

THE PREDICTION DESK

Be less certain.
Be more accurate.

A good forecast is more than a yes or no. Set your probabilities, commit your record, then see how well your confidence matches the outcome.

Your forecast sheet

01—03

These are fictional scenarios with fixed demo outcomes. They are not current market predictions.

THE FEE LEDGER

Accuracy earns
the allocation.

Model an already collected creator-fee budget, after Pons protocol and any buyback allocations. This is a proposed distribution layer, not a connected Pons payout.

Collected creator fees (example units)

Forecast competition · 70%
Next round reserve · 20%
Operations · remainder 10%
Total allocated

The published rule

Brier score = average of (probability − outcome)². Lower is better. A correct outcome is 1; an incorrect outcome is 0.

Reward weight = max(0, 1 − 4 × Brier score). This caps weight at 1 and gives no weight to a score of 0.25 or worse. Divide the competition pool proportionally across eligible weights.

Resolve your sheet to compare with three explicitly simulated competitors.

If every weight is zero, the competition pool carries forward. For this preview, one completed sheet per browser. Real rewards need identity controls, published resolution sources, dispute rules, and a separately audited distributor.

A record worth building.

No points for trading volume. No reward for persuasive prose. Just probabilities committed before a deadline and evaluated against clear outcomes.

The Forecast Exchange

Be less certain.
Be more accurate.

Your forecast sheet

Will the fictional Harbor Observatory record at least 10 mm of rain on Round Day?

Will the fictional Atlas Index close above 1,000 at the round deadline?

Will the fictional North Rail service complete at least 90 of 100 scheduled trips?

A record worth building.

Be less certain.Be more accurate.

Your forecast sheet

Will the fictional Harbor Observatory record at least 10 mm of rain on Round Day?

Will the fictional Atlas Index close above 1,000 at the round deadline?

Will the fictional North Rail service complete at least 90 of 100 scheduled trips?

A record worth building.

Be less certain.
Be more accurate.