Accuracy
How accurate is the Doodles 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)
95.0%
Mean accuracy (100−APE)
92.4%
Within ±5%
50%
Within ±10%
77%
Within ±20%
94%
Bias
+0.5%
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 | 1874 | 95.2% | 80% | +0.9% |
| middle | 28 | 48.3% | 0% | +10.3% |
| grail | 81 | 89.7% | 46% | -11.9% |
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 |
|---|---|---|---|---|---|---|
| #3663 | floor · high | Aug 3, 2026 | 0.381Ξ | 0.421Ξ | +10.5% | 89.5% |
| #5878 | floor · high | Aug 3, 2026 | 0.383Ξ | 0.395Ξ | +3.0% | 97.0% |
| #1861 | floor · high | Aug 3, 2026 | 0.381Ξ | 0.390Ξ | +2.4% | 97.6% |
| #7616 | floor · high | Aug 2, 2026 | 0.439Ξ | 0.396Ξ | -9.8% | 90.2% |
| #2256 | floor · high | Aug 2, 2026 | 0.383Ξ | 0.401Ξ | +4.6% | 95.4% |
| #9154 | floor · high | Aug 2, 2026 | 0.383Ξ | 0.402Ξ | +4.9% | 95.1% |
| #655 | floor · high | Aug 2, 2026 | 0.409Ξ | 0.428Ξ | +4.7% | 95.4% |
| #655 | floor · high | Aug 2, 2026 | 0.401Ξ | 0.453Ξ | +13.0% | 87.0% |
| #7917 | grail · medium | Aug 2, 2026 | 2.600Ξ | 1.227Ξ | -52.8% | 47.2% |
| #100 | floor · high | Aug 2, 2026 | 0.380Ξ | 0.434Ξ | +14.2% | 85.8% |
| #9811 | floor · high | Aug 2, 2026 | 0.381Ξ | 0.434Ξ | +13.9% | 86.1% |
| #8680 | floor · high | Aug 2, 2026 | 0.391Ξ | 0.381Ξ | -2.6% | 97.4% |
| #8680 | floor · high | Aug 2, 2026 | 0.240Ξ | 0.402Ξ | +67.5% | 32.5% |
| #6674 | floor · high | Aug 2, 2026 | 0.471Ξ | 0.495Ξ | +5.1% | 94.9% |
| #5692 | floor · high | Aug 2, 2026 | 0.382Ξ | 0.409Ξ | +7.1% | 92.9% |
| #655 | floor · high | Aug 1, 2026 | 0.449Ξ | 0.431Ξ | -4.0% | 96.0% |
| #6001 | floor · high | Aug 1, 2026 | 0.401Ξ | 0.442Ξ | +10.2% | 89.8% |
| #3146 | floor · high | Aug 1, 2026 | 0.419Ξ | 0.407Ξ | -2.9% | 97.1% |
| #8401 | floor · high | Aug 1, 2026 | 0.420Ξ | 0.407Ξ | -3.1% | 96.9% |
| #6473 | floor · high | Aug 1, 2026 | 0.567Ξ | 0.440Ξ | -22.4% | 77.6% |
| #9656 | floor · high | Jul 31, 2026 | 0.391Ξ | 0.394Ξ | +0.7% | 99.3% |
| #7600 | floor · high | Jul 31, 2026 | 0.383Ξ | 0.388Ξ | +1.2% | 98.8% |
| #8814 | floor · high | Jul 31, 2026 | 0.407Ξ | 0.394Ξ | -3.3% | 96.7% |
| #5878 | floor · high | Jul 31, 2026 | 0.387Ξ | 0.401Ξ | +3.6% | 96.4% |
| #9656 | floor · high | Jul 31, 2026 | 0.386Ξ | 0.400Ξ | +3.6% | 96.4% |
| #7025 | floor · high | Jul 31, 2026 | 0.388Ξ | 0.388Ξ | 0.0% | 100.0% |
| #2412 | floor · high | Jul 31, 2026 | 0.391Ξ | 0.398Ξ | +1.8% | 98.3% |
| #5331 | floor · high | Jul 31, 2026 | 0.390Ξ | 0.398Ξ | +2.1% | 98.0% |
| #3736 | grail · medium | Jul 30, 2026 | 2.299Ξ | 1.994Ξ | -13.3% | 86.7% |
| #5955 | floor · high | Jul 30, 2026 | 0.383Ξ | 0.414Ξ | +8.0% | 92.0% |
| #8814 | floor · high | Jul 30, 2026 | 0.378Ξ | 0.417Ξ | +10.3% | 89.7% |
| #5612 | floor · high | Jul 30, 2026 | 0.488Ξ | 0.448Ξ | -8.2% | 91.8% |
| #3371 | floor · high | Jul 30, 2026 | 0.390Ξ | 0.422Ξ | +8.2% | 91.8% |
| #2202 | floor · high | Jul 30, 2026 | 0.383Ξ | 0.408Ξ | +6.4% | 93.6% |
| #7600 | floor · high | Jul 30, 2026 | 0.380Ξ | 0.421Ξ | +10.8% | 89.2% |
| #5352 | floor · high | Jul 30, 2026 | 0.383Ξ | 0.384Ξ | +0.2% | 99.8% |
| #5747 | floor · high | Jul 30, 2026 | 0.383Ξ | 0.381Ξ | -0.6% | 99.4% |
| #2957 | floor · high | Jul 30, 2026 | 0.383Ξ | 0.385Ξ | +0.4% | 99.6% |
| #244 | floor · high | Jul 30, 2026 | 0.389Ξ | 0.388Ξ | -0.2% | 99.8% |
| #5129 | floor · high | Jul 30, 2026 | 0.390Ξ | 0.407Ξ | +4.4% | 95.6% |
Model doodles-1785728531-2026-08-03.2 · evaluated through 8/3/2026. Method details on the methodology page.