Accuracy
How accurate is the Moonbirds model?
Every score comes from the newest chronological hold-out; the target sale is excluded from training, comparables, and its causal floor proxy.
Median accuracy (100−APE)
94.5%
Mean accuracy (100−APE)
92.3%
Within ±5%
46%
Within ±10%
72%
Within ±20%
93%
Bias
+0.1%
Accuracy by market tier
Floor, middle-market, and grail sales are scored separately so aggregate accuracy cannot hide a weak segment
| Tier | Sales | Median accuracy | Within ±10% | Bias |
|---|---|---|---|---|
| floor | 1495 | 94.5% | 72% | +0.3% |
| middle | 5 | 89.9% | 40% | -17.4% |
| grail | 13 | 80.8% | 15% | -16.8% |
Weekly median accuracy
Chronological hold-out accuracy by sale week
Error distribution
Signed hold-out error; positive means over-predicted
Scored hold-out sales
60 out-of-sample predictions frozen at evaluation time
| Item | Tier / evidence | Date | Sale | Model | Error | Accuracy |
|---|---|---|---|---|---|---|
| #3790 | grail · high | Aug 5, 2026 | 5.250Ξ | 2.831Ξ | -46.1% | 53.9% |
| #4364 | floor · high | Aug 4, 2026 | 0.545Ξ | 0.520Ξ | -4.6% | 95.4% |
| #2311 | floor · high | Aug 4, 2026 | 0.528Ξ | 0.618Ξ | +17.1% | 83.0% |
| #3222 | floor · high | Aug 4, 2026 | 0.539Ξ | 0.518Ξ | -3.9% | 96.1% |
| #7529 | floor · high | Aug 4, 2026 | 0.539Ξ | 0.519Ξ | -3.7% | 96.3% |
| #6703 | floor · high | Aug 4, 2026 | 0.511Ξ | 0.564Ξ | +10.4% | 89.6% |
| #3222 | floor · high | Aug 4, 2026 | 0.505Ξ | 0.574Ξ | +13.7% | 86.3% |
| #1481 | floor · high | Aug 4, 2026 | 0.550Ξ | 0.567Ξ | +3.1% | 96.9% |
| #1354 | floor · high | Aug 4, 2026 | 0.531Ξ | 0.572Ξ | +7.7% | 92.3% |
| #5210 | floor · high | Aug 4, 2026 | 0.540Ξ | 0.575Ξ | +6.5% | 93.5% |
| #8910 | floor · high | Aug 4, 2026 | 0.540Ξ | 0.588Ξ | +8.9% | 91.1% |
| #1871 | floor · high | Aug 4, 2026 | 0.582Ξ | 0.592Ξ | +1.8% | 98.2% |
| #7453 | floor · high | Aug 3, 2026 | 0.544Ξ | 0.604Ξ | +11.0% | 89.0% |
| #7807 | floor · high | Aug 3, 2026 | 0.550Ξ | 0.593Ξ | +7.8% | 92.2% |
| #5661 | floor · high | Aug 2, 2026 | 0.543Ξ | 0.631Ξ | +16.2% | 83.8% |
| #5369 | floor · high | Aug 2, 2026 | 0.600Ξ | 0.631Ξ | +5.2% | 94.8% |
| #4364 | floor · high | Aug 2, 2026 | 0.560Ξ | 0.605Ξ | +8.0% | 92.0% |
| #703 | floor · high | Aug 1, 2026 | 0.750Ξ | 0.611Ξ | -18.5% | 81.5% |
| #5429 | floor · high | Aug 1, 2026 | 0.642Ξ | 0.757Ξ | +17.9% | 82.1% |
| #5314 | floor · high | Aug 1, 2026 | 0.630Ξ | 0.613Ξ | -2.7% | 97.3% |
| #299 | floor · high | Aug 1, 2026 | 0.590Ξ | 0.600Ξ | +1.8% | 98.2% |
| #6247 | floor · high | Aug 1, 2026 | 0.589Ξ | 0.606Ξ | +2.8% | 97.2% |
| #8456 | floor · high | Aug 1, 2026 | 0.568Ξ | 0.626Ξ | +10.2% | 89.8% |
| #6760 | floor · high | Aug 1, 2026 | 0.576Ξ | 0.582Ξ | +1.0% | 99.0% |
| #8193 | floor · high | Jul 31, 2026 | 0.574Ξ | 0.573Ξ | -0.1% | 99.9% |
| #6746 | floor · high | Jul 31, 2026 | 0.569Ξ | 0.605Ξ | +6.3% | 93.7% |
| #9080 | floor · high | Jul 31, 2026 | 0.573Ξ | 0.589Ξ | +2.7% | 97.3% |
| #9892 | floor · high | Jul 31, 2026 | 0.570Ξ | 0.599Ξ | +5.1% | 94.9% |
| #9854 | floor · high | Jul 31, 2026 | 0.574Ξ | 0.614Ξ | +7.0% | 93.0% |
| #8165 | floor · high | Jul 30, 2026 | 0.582Ξ | 0.626Ξ | +7.6% | 92.4% |
| #2977 | floor · high | Jul 30, 2026 | 0.583Ξ | 0.605Ξ | +3.8% | 96.2% |
| #2944 | floor · high | Jul 30, 2026 | 0.629Ξ | 0.594Ξ | -5.6% | 94.4% |
| #5963 | floor · high | Jul 30, 2026 | 0.605Ξ | 0.591Ξ | -2.3% | 97.7% |
| #31 | floor · high | Jul 30, 2026 | 0.598Ξ | 0.661Ξ | +10.5% | 89.5% |
| #891 | floor · high | Jul 30, 2026 | 0.650Ξ | 0.617Ξ | -5.1% | 94.9% |
| #8387 | floor · high | Jul 30, 2026 | 0.577Ξ | 0.623Ξ | +8.0% | 92.0% |
| #1171 | floor · high | Jul 30, 2026 | 0.629Ξ | 0.596Ξ | -5.3% | 94.8% |
| #5853 | floor · high | Jul 29, 2026 | 0.580Ξ | 0.617Ξ | +6.4% | 93.6% |
| #8755 | floor · high | Jul 29, 2026 | 0.590Ξ | 0.641Ξ | +8.6% | 91.4% |
| #547 | floor · high | Jul 29, 2026 | 0.640Ξ | 0.658Ξ | +2.8% | 97.2% |
Model moonbirds-1785890687-2026-08-03.2 · evaluated through 8/5/2026. Method details on the methodology page.