Phantom
High-conviction: only commits when Elo + WR10 strongly agree.
All-time accuracy
66.7%
6 / 9 settled
Last 50
66.7%
sliding window
Last 10
66.7%
most recent
Brier score
0.214
lower = better (0.25 = coin flip)
Log-loss
0.618
lower = better
Simulated bankroll
If you'd staked $10 on every prediction at fair odds (1/probability, no bookie vig), your P&L would be +$1.59 across 9 bets. ROI = +1.77%.
Calibration by confidence
When the model says 80%, is it actually right 80% of the time? Well-calibrated models have accuracy that matches confidence range.
| Confidence range | Picks | Accuracy | Calibration |
|---|---|---|---|
| 50–60% | 0 | — | |
| 60–70% | 8 | 62.5% (5/8) | |
| 70–80% | 1 | 100.0% (1/1) | |
| 80–90% | 0 | — | |
| 90–101% | 0 | — |
Upcoming picks
| When | Match | This model picks | Confidence |
|---|---|---|---|
| GL vs XG | GL 79% | 0.20 |
Recent picks (10)
| When | Match | Pick | Outcome | Result |
|---|---|---|---|---|
| Jul 01 | GL vs XG | GL 79% | pending | — |
| Jun 27 | Resilience vs YB | Resilience 71% | Resilience | ✓ HIT |
| Jun 27 | LGD vs PlayTime | LGD 65% | LGD | ✓ HIT |
| Jun 27 | LGD vs PlayTime | LGD 65% | LGD | ✓ HIT |
| Jun 27 | LGD vs PlayTime | LGD 65% | LGD | ✓ HIT |
| Jun 27 | NMSS vs AMG | NMSS 64% | AMG | ✗ miss |
| Jun 27 | NMSS vs AMG | NMSS 64% | AMG | ✗ miss |
| Jun 27 | NMSS vs AMG | NMSS 64% | AMG | ✗ miss |
| Jun 27 | LGD vs PlayTime | LGD 65% | LGD | ✓ HIT |
| Jun 27 | LGD vs PlayTime | LGD 65% | LGD | ✓ HIT |