Accuracy
How accurate is the mfers 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)
93.6%
Mean accuracy (100−APE)
90.8%
Within ±5%
39%
Within ±10%
68%
Within ±20%
91%
Bias
-1.7%
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 | 1439 | 93.7% | 69% | -1.4% |
| middle | 27 | 76.0% | 22% | -2.3% |
| grail | 21 | 85.1% | 33% | -19.0% |
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 |
|---|---|---|---|---|---|---|
| #694 | floor · high | Aug 2, 2026 | 0.230Ξ | 0.199Ξ | -13.5% | 86.5% |
| #9956 | floor · high | Aug 2, 2026 | 0.193Ξ | 0.216Ξ | +11.9% | 88.1% |
| #6775 | floor · high | Aug 1, 2026 | 0.194Ξ | 0.200Ξ | +3.1% | 96.9% |
| #6865 | floor · high | Aug 1, 2026 | 0.232Ξ | 0.226Ξ | -2.6% | 97.4% |
| #4346 | floor · high | Aug 1, 2026 | 0.220Ξ | 0.198Ξ | -10.0% | 90.0% |
| #1243 | floor · high | Aug 1, 2026 | 0.183Ξ | 0.200Ξ | +9.3% | 90.7% |
| #2294 | floor · high | Aug 1, 2026 | 0.188Ξ | 0.194Ξ | +3.2% | 96.8% |
| #253 | floor · high | Aug 1, 2026 | 0.188Ξ | 0.195Ξ | +3.7% | 96.3% |
| #8239 | floor · high | Aug 1, 2026 | 0.189Ξ | 0.194Ξ | +2.7% | 97.4% |
| #6775 | floor · high | Aug 1, 2026 | 0.200Ξ | 0.189Ξ | -5.5% | 94.5% |
| #4346 | floor · high | Aug 1, 2026 | 0.185Ξ | 0.218Ξ | +17.8% | 82.2% |
| #2907 | floor · high | Aug 1, 2026 | 0.250Ξ | 0.200Ξ | -20.0% | 80.0% |
| #2079 | floor · high | Jul 31, 2026 | 0.196Ξ | 0.190Ξ | -2.9% | 97.1% |
| #6912 | floor · high | Jul 31, 2026 | 0.199Ξ | 0.187Ξ | -5.9% | 94.1% |
| #1217 | floor · high | Jul 31, 2026 | 0.199Ξ | 0.192Ξ | -3.4% | 96.6% |
| #1217 | floor · high | Jul 31, 2026 | 0.188Ξ | 0.197Ξ | +4.8% | 95.2% |
| #6775 | floor · high | Jul 31, 2026 | 0.188Ξ | 0.194Ξ | +3.2% | 96.8% |
| #2079 | floor · high | Jul 31, 2026 | 0.188Ξ | 0.204Ξ | +8.5% | 91.5% |
| #5789 | floor · high | Jul 31, 2026 | 0.187Ξ | 0.262Ξ | +40.1% | 59.9% |
| #255 | floor · high | Jul 31, 2026 | 0.200Ξ | 0.200Ξ | 0.0% | 100.0% |
| #2079 | floor · high | Jul 31, 2026 | 0.200Ξ | 0.188Ξ | -5.9% | 94.1% |
| #1243 | floor · high | Jul 31, 2026 | 0.200Ξ | 0.192Ξ | -3.8% | 96.2% |
| #6775 | floor · high | Jul 31, 2026 | 0.199Ξ | 0.193Ξ | -3.3% | 96.7% |
| #2294 | floor · high | Jul 31, 2026 | 0.199Ξ | 0.192Ξ | -3.8% | 96.2% |
| #253 | floor · high | Jul 30, 2026 | 0.194Ξ | 0.187Ξ | -3.6% | 96.4% |
| #8239 | floor · high | Jul 30, 2026 | 0.194Ξ | 0.198Ξ | +2.1% | 97.9% |
| #2294 | floor · high | Jul 30, 2026 | 0.186Ξ | 0.216Ξ | +16.1% | 83.9% |
| #3154 | floor · high | Jul 30, 2026 | 0.200Ξ | 0.195Ξ | -2.5% | 97.5% |
| #3154 | floor · high | Jul 30, 2026 | 0.187Ξ | 0.199Ξ | +6.4% | 93.6% |
| #808 | floor · high | Jul 30, 2026 | 0.200Ξ | 0.192Ξ | -3.9% | 96.1% |
| #694 | floor · high | Jul 30, 2026 | 0.195Ξ | 0.189Ξ | -3.0% | 97.0% |
| #7137 | floor · high | Jul 30, 2026 | 0.186Ξ | 0.195Ξ | +4.8% | 95.2% |
| #7283 | floor · high | Jul 30, 2026 | 0.186Ξ | 0.230Ξ | +23.7% | 76.3% |
| #253 | floor · high | Jul 30, 2026 | 0.187Ξ | 0.256Ξ | +36.9% | 63.1% |
| #808 | floor · high | Jul 30, 2026 | 0.188Ξ | 0.219Ξ | +16.5% | 83.5% |
| #694 | floor · high | Jul 30, 2026 | 0.192Ξ | 0.225Ξ | +17.2% | 82.8% |
| #1586 | floor · high | Jul 29, 2026 | 0.208Ξ | 0.205Ξ | -1.3% | 98.7% |
| #1260 | floor · high | Jul 28, 2026 | 0.211Ξ | 0.203Ξ | -3.8% | 96.2% |
| #8720 | floor · high | Jul 28, 2026 | 0.211Ξ | 0.200Ξ | -5.3% | 94.8% |
| #70 | floor · high | Jul 28, 2026 | 0.211Ξ | 0.209Ξ | -1.0% | 99.0% |
Model mfers-1785642719-2026-08-03.2 · evaluated through 8/2/2026. Method details on the methodology page.