Plate Base

Methodology

How Plate Base values UK number plates

Every Plate Base valuation is the output of a stack of independent models, blended together with a confidence band. Nothing is hand-priced. This page documents exactly what feeds each valuation and how the layers interact.

Try it on a plate See our data sources

The core layers

  1. Structural layer — format (dateless, prefix, suffix, current), length and character positions establish a rarity baseline.
  2. Visual Plate Recognition Score (VPRS) — decodes leet substitutions (5→S, 1→I, 0→O, 8→B, 6/9→G) and fuzzy-matches the plate against dictionaries of names, brands, and common words.
  3. Automotive layer — brand, model and engine-code recognition (R8, GT3, 911, M3, AMG, RS6) with prestige floors.
  4. Iconic / trophy anchors — grail plates (F1, A1, S1, 1D, etc.) are anchored to their true collector value.
  5. Prestige dateless structure — short dateless plates receive a dedicated rarity band multiplier.
  6. Community & demand layers — initials, popular first names, cultural and religious references.
  7. Liquidity model — outputs a price ladder and a plain-English liquidity score.

The comparables engine

On top of the structural layers, Plate Base runs a live comparables search across the major UK marketplaces every day. Results feed the plate_market_data store which the valuation engine reads on every request. Comparables are decoded via VPRS so K1 NGS and KIN 95 correctly surface each other as “KINGS”.

How we produce the final range

The final estimate is the median of the surviving models, weighted by reliability tier. We report a low, likely and high band; the width of the band reflects the confidence the models have in each other. Iconic and trophy anchors override the range when a plate is recognised as a “grail”.

What we deliberately do NOT do

  • We do not return the DVLA reserve price as a valuation.
  • We do not average random listings that happen to share characters.
  • We do not sell your data or use valuations to generate cold-call leads.

Editorial oversight

Model outputs are reviewed weekly against realised sales. Iconic anchors, prestige bands and the market-data store are versioned and auditable. Any material change to a model is reflected in the confidence band, not hidden.