Accuracy
How accurate is the Otherdeed for Otherside 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)
93.8%
Mean accuracy (100−APE)
86.7%
Within ±5%
44%
Within ±10%
60%
Within ±20%
81%
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 | 1668 | 94.7% | 65% | +0.1% |
| middle | 58 | 60.5% | 9% | -4.0% |
| grail | 277 | 88.3% | 44% | +4.3% |
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 |
|---|---|---|---|---|---|---|
| #27274 | floor · high | Sep 17, 2026 | 0.068Ξ | 0.080Ξ | +18.3% | 81.7% |
| #48089 | floor · high | Sep 17, 2026 | 0.068Ξ | 0.082Ξ | +21.3% | 78.7% |
| #11356 | floor · high | Sep 17, 2026 | 0.066Ξ | 0.080Ξ | +20.9% | 79.2% |
| #28670 | grail · high | Sep 17, 2026 | 0.068Ξ | 0.082Ξ | +21.3% | 78.7% |
| #30593 | floor · high | Sep 17, 2026 | 0.068Ξ | 0.079Ξ | +16.9% | 83.1% |
| #55257 | floor · high | Sep 17, 2026 | 0.066Ξ | 0.079Ξ | +19.7% | 80.3% |
| #54345 | floor · high | Sep 17, 2026 | 0.066Ξ | 0.078Ξ | +18.0% | 82.0% |
| #47393 | floor · high | Sep 16, 2026 | 0.078Ξ | 0.083Ξ | +6.1% | 93.9% |
| #46075 | middle · high | Sep 16, 2026 | 0.073Ξ | 0.182Ξ | +149.7% | 0.0% |
| #46117 | floor · high | Sep 16, 2026 | 0.100Ξ | 0.081Ξ | -19.0% | 81.0% |
| #57734 | floor · high | Sep 16, 2026 | 0.148Ξ | 0.075Ξ | -49.3% | 50.7% |
| #83262 | floor · high | Sep 15, 2026 | 0.095Ξ | 0.077Ξ | -18.6% | 81.4% |
| #10339 | floor · high | Sep 15, 2026 | 0.080Ξ | 0.079Ξ | -0.9% | 99.1% |
| #88345 | floor · high | Sep 15, 2026 | 0.080Ξ | 0.080Ξ | +0.5% | 99.5% |
| #47375 | floor · high | Sep 15, 2026 | 0.073Ξ | 0.077Ξ | +5.9% | 94.1% |
| #55508 | grail · high | Sep 14, 2026 | 0.440Ξ | 0.108Ξ | -75.5% | 24.6% |
| #57734 | floor · high | Sep 14, 2026 | 0.071Ξ | 0.107Ξ | +50.9% | 49.1% |
| #94135 | grail · high | Sep 14, 2026 | 0.073Ξ | 0.080Ξ | +9.3% | 90.7% |
| #2381 | floor · high | Sep 14, 2026 | 0.189Ξ | 0.094Ξ | -50.3% | 49.7% |
| #50153 | floor · high | Sep 13, 2026 | 0.074Ξ | 0.077Ξ | +3.4% | 96.6% |
| #9928 | floor · high | Sep 13, 2026 | 0.121Ξ | 0.084Ξ | -30.6% | 69.4% |
| #55196 | floor · high | Sep 12, 2026 | 0.094Ξ | 0.075Ξ | -20.4% | 79.6% |
| #68576 | floor · high | Sep 12, 2026 | 0.083Ξ | 0.075Ξ | -9.6% | 90.4% |
| #6984 | middle · high | Sep 12, 2026 | 0.243Ξ | 0.150Ξ | -38.4% | 61.6% |
| #88855 | floor · high | Sep 11, 2026 | 0.090Ξ | 0.086Ξ | -4.3% | 95.7% |
| #2265 | floor · high | Sep 11, 2026 | 0.165Ξ | 0.109Ξ | -33.9% | 66.1% |
| #85844 | floor · high | Sep 11, 2026 | 0.080Ξ | 0.072Ξ | -9.7% | 90.3% |
| #76212 | floor · high | Sep 10, 2026 | 0.085Ξ | 0.077Ξ | -9.0% | 91.0% |
| #45809 | floor · high | Sep 10, 2026 | 0.083Ξ | 0.074Ξ | -10.4% | 89.6% |
| #47393 | floor · high | Sep 10, 2026 | 0.083Ξ | 0.075Ξ | -9.3% | 90.7% |
| #88855 | floor · high | Sep 10, 2026 | 0.086Ξ | 0.071Ξ | -17.9% | 82.1% |
| #89125 | middle · high | Sep 10, 2026 | 0.533Ξ | 0.103Ξ | -80.7% | 19.3% |
| #3173 | middle · high | Sep 10, 2026 | 0.185Ξ | 0.129Ξ | -30.4% | 69.7% |
| #66262 | floor · high | Sep 10, 2026 | 0.100Ξ | 0.068Ξ | -32.0% | 68.0% |
| #63714 | floor · high | Sep 10, 2026 | 0.103Ξ | 0.077Ξ | -25.2% | 74.8% |
| #16250 | floor · high | Sep 10, 2026 | 0.093Ξ | 0.084Ξ | -9.7% | 90.3% |
| #19497 | floor · high | Sep 10, 2026 | 0.094Ξ | 0.083Ξ | -11.6% | 88.4% |
| #19497 | floor · high | Sep 10, 2026 | 0.082Ξ | 0.084Ξ | +2.3% | 97.7% |
| #54795 | grail · insufficient | Sep 10, 2026 | 1.350Ξ | 0.648Ξ | -52.0% | 48.0% |
| #63017 | grail · low | Sep 10, 2026 | 0.990Ξ | 0.375Ξ | -62.1% | 37.9% |
Model otherdeed-1789609439-2026-08-03.2 · evaluated through 9/17/2026. Method details on the methodology page.