When this trader buys at $0.70, they imply 70% probability. Perfect calibration = the event happens 70% of the time.
| Bucket | Bets | Expected | Actual | Error |
|---|---|---|---|---|
| 0.10-0.20 | 15 | 14% | 0% | 14.4% |
| 0.20-0.30 | 6 | 26% | 0% | 26.4% |
| 0.30-0.40 | 18 | 34% | 0% | 34.0% |
| 0.40-0.50 | 46 | 45% | 0% | 45.5% |
| 0.50-0.60 | 28 | 54% | 4% | 50.3% |
| 0.60-0.70 | 22 | 65% | 0% | 64.6% |
| 0.70-0.80 | 22 | 75% | 0% | 74.9% |
| 0.80-0.90 | 13 | 83% | 0% | 83.1% |
| 0.90-1.00 | 3 | 94% | 0% | 93.6% |
On-chain verification: wallet age 105 days, 2899 txs, provenance grade B. Bot score: 30/100, wash trading score: 20/100.
Polymarket on-chain coverage: $202 in / $0 out across 14 withdrawal tx since 2026-02-03.
3100 total trades across 216 markets.
173 bets on resolved markets available for calibration scoring.
Calibration error: 51.5% — needs improvement.
Skill: 0/100 (calibration quality). Variance: 22/100 (higher = more volatile returns).
Brier Skill Score: -5300.3% vs naive baseline (>0% = better than always predicting base rate).
Brier decomposition: REL=0.3036 RES=0.0002 UNC=0.0057.
Log loss: 0.8470 (skill: -2282.6% vs naive). Lower log loss = better calibration on rare events.
Timeliness: avg entry 0.5 days before resolution, 0% early mover.
Below average. The data shows poor calibration, thin evidence, or both. When this trader expresses high confidence, events don't happen at the rate they imply.
Confidence: D/38 [CI95: D→D, 36-40] (173 resolved bets). Moderate confidence — score may shift as more markets resolve.
Methodology: Brier Score Decomposition (Murphy 1973), Log Loss, On-Chain USDC Verification. Same approach used by IARPA to identify superforecasters.