Platform Accuracy
How well do OddsForge markets predict real-world outcomes? When a market says an event has X% chance, does it actually happen X% of the time?
Loading accuracy data...
Accuracy data is temporarily unavailable
—
—
Lower is better (0 = perfect)
—
Percentage of markets resolved YES
—
Calibration Chart
Predicted probability vs. actual outcome frequency. A perfectly calibrated market follows the diagonal line.
Monthly Resolutions
Category Breakdown
Calibration Breakdown
| Price Bucket | Markets | Resolved YES | Expected Rate | Actual Rate | Deviation |
|---|
Recent Resolutions
| Market | Outcome | Final Yes Price | Category | Resolved |
|---|
Methodology
Calibration Chart: Markets are grouped into 10 probability buckets (0–10%, 10–20%, etc.) based on their YES price at the time of resolution. For each bucket, we calculate the actual fraction that resolved YES. A perfectly calibrated platform would have points falling exactly on the diagonal — meaning when the market says 70%, the event happens 70% of the time.
Brier Score: A standard measure of forecast accuracy. It's the mean squared error between the predicted probability (YES price) and the actual outcome (1 for YES, 0 for NO). A score of 0 is perfect, 0.25 means random guessing (coin flip), and 1.0 is always wrong.
Directional Accuracy: The percentage of non-toss-up markets (YES price ≠ 50¢) where the market correctly predicted the direction of the outcome — i.e., price > 50¢ and resolved YES, or price < 50¢ and resolved NO.