AppraisrAppraisr
Accuracy

How accurate is the Azuki Elementals 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)
88.2%
Mean accuracy (100−APE)
82.3%
Within ±5%
26%
Within ±10%
45%
Within ±20%
72%
Bias
+7.8%

Accuracy by market tier

Floor, middle-market, and grail sales are scored separately so aggregate accuracy cannot hide a weak segment

TierSalesMedian accuracyWithin ±10%Bias
floor211889.2%48%+5.9%
middle11970.5%13%+3.2%
grail12172.4%29%+46.0%

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

ItemTypeTier / evidenceDateSaleModelErrorAccuracy
#1645Humanfloor · highOct 8, 20260.120Ξ0.076Ξ-36.7%63.3%
#17129Humanfloor · highOct 8, 20260.057Ξ0.066Ξ+15.6%84.4%
#4611Humanfloor · highOct 8, 20260.061Ξ0.063Ξ+3.7%96.3%
#10145Humanfloor · highOct 8, 20260.076Ξ0.062Ξ-18.5%81.5%
#4611Humanfloor · highOct 8, 20260.057Ξ0.069Ξ+20.8%79.2%
#1645Humanfloor · highOct 7, 20260.075Ξ0.067Ξ-10.7%89.3%
#4267Redfloor · insufficientOct 7, 20260.058Ξ0.075Ξ+29.1%70.9%
#295Humanfloor · highOct 7, 20260.061Ξ0.067Ξ+10.3%89.7%
#295Humanfloor · highOct 7, 20260.057Ξ0.079Ξ+38.6%61.4%
#18505Humanfloor · highOct 7, 20260.057Ξ0.072Ξ+25.9%74.1%
#19625Humanmiddle · highOct 7, 20260.107Ξ0.106Ξ-0.9%99.1%
#10145Humanfloor · highOct 7, 20260.058Ξ0.067Ξ+16.5%83.5%
#18783Animalgrail · highOct 6, 20260.205Ξ0.295Ξ+43.9%56.1%
#5269Humanfloor · highOct 6, 20260.079Ξ0.064Ξ-18.5%81.5%
#12961Humanfloor · highOct 6, 20260.058Ξ0.060Ξ+2.7%97.3%
#1877Humanfloor · highOct 6, 20260.058Ξ0.062Ξ+6.7%93.3%
#992Humanfloor · highOct 6, 20260.058Ξ0.060Ξ+3.2%96.8%
#9255Humanfloor · highOct 6, 20260.058Ξ0.066Ξ+14.8%85.2%
#12961Humanfloor · highOct 6, 20260.056Ξ0.069Ξ+22.3%77.7%
#9933Humanfloor · highOct 6, 20260.058Ξ0.080Ξ+39.1%60.9%
#1877Humanfloor · highOct 6, 20260.057Ξ0.067Ξ+17.5%82.5%
#992Humanfloor · highOct 6, 20260.057Ξ0.063Ξ+11.1%88.9%
#5228Humanfloor · highOct 6, 20260.061Ξ0.063Ξ+2.8%97.2%
#4656Humanfloor · highOct 6, 20260.065Ξ0.060Ξ-7.0%93.0%
#5228Humanfloor · highOct 6, 20260.057Ξ0.063Ξ+10.3%89.7%
#14310Humanmiddle · highOct 6, 20260.095Ξ0.153Ξ+61.1%39.0%
#18159Humanfloor · highOct 5, 20260.061Ξ0.061Ξ-0.5%99.5%
#3848Humanfloor · highOct 5, 20260.061Ξ0.062Ξ+1.1%98.9%
#3848Humanfloor · highOct 5, 20260.057Ξ0.069Ξ+20.6%79.4%
#4139Humanfloor · highOct 5, 20260.059Ξ0.062Ξ+5.5%94.5%
#11859Humanfloor · highOct 5, 20260.059Ξ0.061Ξ+3.8%96.2%
#5534Humanfloor · highOct 5, 20260.069Ξ0.063Ξ-8.1%91.9%
#11859Humanfloor · highOct 5, 20260.057Ξ0.067Ξ+17.5%82.5%
#17448Humanfloor · highOct 5, 20260.059Ξ0.059Ξ+0.4%99.6%
#1056Humanfloor · highOct 5, 20260.059Ξ0.065Ξ+10.6%89.4%
#8931Humanfloor · highOct 5, 20260.059Ξ0.061Ξ+3.8%96.2%
#14675Humanfloor · highOct 5, 20260.059Ξ0.061Ξ+2.6%97.4%
#9396Humanfloor · highOct 5, 20260.055Ξ0.067Ξ+20.9%79.1%
#5269Humanfloor · highOct 5, 20260.059Ξ0.073Ξ+24.8%75.2%
#4656Humanfloor · highOct 5, 20260.059Ξ0.064Ξ+9.4%90.6%

Model azuki-elementals-1791483287-2026-08-03.2 · evaluated through 10/8/2026. Method details on the methodology page.