OTHERDEED EXPANDED#758

Sales-guided estimate · saved allocation method
rare-t5-consistency-2026-09-08 · calculated 2026-09-21

Standard Luster land: 0.2341 ETH, a weighted-median reference from 6 plain and 0 supported stacked deeds. This parcel's sediment/common-trait adjustment is ×1.5588: 0.365 ETH land.

At this build: 0.365 land + 0.0489 package difference + 0 tier/resource ordering adjustment = 0.4138 ETH. A package difference is a conditional allocation, not an independently proven resource price.

Historical normalization: sale ETH × 0.0861 / sale-date sales reference, applied once. The model uses the available history within the current 180-day search; recent sales receive more weight. Repeated sales of the same deed supply one observation.

Matching package evidence · 15 distinct deeds.

Land evidence. The last column is each receipt's conditional plain-land equivalent before this parcel's common-trait adjustment. The active land reference is the weighted median of those adjusted observations. It is not the highest residual or a sum of the rows.

Earlier matching-resource evidence supplies 90% of the extras allowance; the existing component model supplies the remainder. The active blend retains its existing effective-sample and disagreement weights; the candidate absolute-age revision is still under validation.

Between builds: each candidate follows fixed allowance + market-sensitive allowance × live reference / build reference; the highest candidate sets the point. At this build the winning candidate contains 0 ETH fixed and 0.4138 ETH market-sensitive value. Sale-derived package allowances follow the sales reference; inherited fixed components remain fixed.

Broader stacked-sale land fitting is under validation. Artifact and deeper-resource receipts remain useful package evidence, but the proposed additional plain-land deductions are not active: historical checks were mixed. This release preserves the existing land evidence and corrects how its component amounts travel through market changes and Ask.

The references are reconstructed from sales, not historical asking floors. Component deductions use earlier matching sales and the current lower-tier trait model. Sparse comparisons and unexplained package premiums remain uncertain; no calibrated range is published. Full method and limitations