Back to the map

How Posh Maps works

Posh Maps is a cartographic publication, not a valuation service. Every number on it is derived from open data with a method we explain in full here. Where a figure is an estimate, we say so — and we show how confident we are.

The estimated value (AVM)

We do not use a black-box machine-learning model. The estimate is a transparent formula you can check by hand. For any property with a recorded sale:

estimate = last_sold_price
         × HPI_multiplier        (borough house-price index, then ÷ now)
         × type_adjustment       (flat 0.95 · terraced 1.00 · semi 1.02 · detached 1.05)
         × new_build_decay        (a 2019 new-build is now a 7-year-old flat)

The further back the last sale, the wider the confidence band: under 2 years ±5%, under 5 years ±10%, under 10 years ±15%, otherwise ±25%. Backtested against repeat sales in the last 12 months, the median absolute error is under 15%.

Which sale we use. Where a building on the map carries its own address, we use the latest sale at that address; otherwise the nearest sale in its postcode. Either way we skip sales that are probably not full-market prices — shared-ownership shares and discounted sales, which the Land Registry records without a flag. A sale is set aside if the same home resold within 3 years for more than 2.2× (after house-price inflation); if it was under 0.45× the typical price of the same type of home in its postcode around that time; if it is a new-build flat priced at a round share (25–40%) of its block, alongside several others at the same share; or if the address names a housing association. About 0.7% of London sales are set aside. They still appear on the map, marked “may not be a full-market sale”.

The gentrification score

This is a heuristic, not a forecast. It aggregates observable proxies that correlate with neighbourhood change, weights them transparently, and shows you the inputs. Per LSOA (≈1,500 residents), five components in [0,1]:

A momentum cluster (price growth + new development + turnover, ~0.64 of the weight) captures areas changing right now — Wembley Park, Nine Elms, Barking Riverside — which amenity signals lag. The rest nudge established areas up.

The score describes an area as it is now; it deliberately conflates “currently gentrifying” with “already gentrified”. We do not surface demographic, ethnic, or income data — the other components do the job without the ethical hazard.

Your priorities

Rate ten things from 0 (doesn’t matter) to 5 (must have): low crime, good schools, transport, parks, peace and quiet, nightlife and restaurants, shops and cafés, value for money, an upmarket feel and clean air. Each is a London rank from 0 to 100, so 70 means better than 70% of London. An area’s score is the average of those ranks, weighted by your ratings. A must-have the area falls below London’s middle on costs points: 0.6 per point short, so a must-have at 20/100 takes 18 off. Click any area to see how its score adds up, line by line.

The Overall area score

Where gentrification measures change, the Overall area score measures current desirability, with weights you can change. Six factors are computed per LSOA, eachpercentile-normalised 0–100 across London (so 70 means “better than 70% of London”), then combined using importance weights you set 1–5 in the sidebar:

Drag the 1–5 importance sliders (or pick a preset — Young professional, Family, Quiet/leafy, Investor) and the map + popups recompute live to your definition of a good area; the popup shows the per-factor breakdown so the number is explainable. Where a source is sparse (e.g. no school within 1.2 km) the factor falls back to a neutral 50.

Affluence is price-derived by design; it correlates with area price-per-m² at r ≈ 0.81, confirming “high-end” tracks real area wealth. (Census occupation blending is a future refinement.)

The Posh score

In one line: how upmarket a neighbourhood is, from what homes are worth per m², the shops nearby, the high street, home ownership, crime, schools, parks and connections. It is not a measure of how good a place is to live: for that, use Your priorities or Find your area.

The Posh score (and How posh is your postcode?) asks one question: how upmarket is this neighbourhood? It uses fixed weights, so everyone sees the same number. Each factor is ranked against all 5,042 neighbourhoods we cover (Greater London plus 48 in Brentwood, Essex) (0–100), and the score is the weighted sum, so the points in a breakdown add up to the score.

A postcode gets the score of the neighbourhood (LSOA, about 1,500 residents) it sits in. A district such as SW3 is the average of its neighbourhoods, weighted by how many postcodes each holds. Bands: under 20 Ordinary, 20–40 Emerging, 40–60 Comfortable, 60–80 Upmarket, 80+ Exclusive.

Price growth (real terms)

The growth map shows how each area’s typical sale price changed over a window you choose. We precompute the median sale price per LSOA per year (1995–today) from the full transaction history, smooth each endpoint over a 3-year band, and compare them. Areas with too few sales in either band are flagged and left off the map rather than shown as noise.

By default we report real change — deflated to the start-year’s money using the ONS CPIH index (annual average, 2015 = 100; swappable for the monthly series) — so “growth” means real value, not just a falling pound. A nominal toggle shows headline £ change. Over 2011→2026, for example, the typical London area rose ~+79% nominal but only ~+24% real; much of prime central London is down in real terms. The diverging scale is centred at 0% real change.

Find your area

For your spec (type, bedrooms, budget) we show, per area, the share of comparable homes whose estimated value is at or under your budget — e.g. “82% of 2-bed flats here are under £600k.” Greener areas have more options within reach. (Earlier this showed a £/m² ratio, which was both unintuitive and noisy.)

Legacy note — the original £/m² method:

We translate your budget into a price-per-m² using a typical floor area for the property type, then compare it to the median price-per-m² of comparable homes in each area. Greener areas transact below your budget; redder areas above it.

The mover engine

Add a commute anchor (station or postcode) and a few priorities, and Find-for-me becomes an answer engine. We build a commute isochrone — the areas reachable within your time limit by public transport — then rank the areas inside it by fit, a blend of affordability (share of homes under budget) and your personalised Area Score. Each result carries one honest trade-off (“the catch”), so it reads like advice, not a brochure. Commute times are real door-to-door journeys from the centre of each area to your destination, arriving at 9am on a weekday (TravelTime timetabled routing), cached per destination for 90 days. Against TfL’s Journey Planner they are typically within 4 minutes. If routing is unavailable we approximate from the station network and label times “approximate”.

Area verdicts are short, plain-English summaries generated only from an area’s real signals (price, factor scores, decade trajectory, who it suits, the weakest point). They’re precomputed and cached per area — never invented per pageview — and regenerate on data refresh. With an OpenRouter key they’re written by a language model from those signals; otherwise a deterministic template assembles them from the same data.

Estimated rent

No free dataset gives per-property rents, so we anchor to ONS Private Rental Market median rents (by borough) and scale each home by its value relative to its borough median — a per-borough gross-yield applied to the estimate. It’s a model (shown as such), but it gives rent at the same granularity as the price map. The rental spec (flat/house, bedrooms, bathrooms) applies London-wide bedroom/type/bath multipliers from ONS rent-by- bedroom ratios.

Data sources

All open data, under the Open Government Licence v3.0 unless noted.

Contains HM Land Registry data © Crown copyright and database right 2026, licensed under the Open Government Licence v3.0. Contains public sector information licensed under the OGL v3.0. Basemap © CARTO, © OpenStreetMap contributors.

Known limitations

Where a title polygon could not be matched to an address, pricing falls back to the nearest recorded sale in the same postcode — accurate at postcode level, approximate at the individual building. Floor area is from EPC where available, otherwise modelled from the property type (shown as “modelled”). Shared-ownership sales can pull a few areas’ price-per-m² down. Loading the borough’s INSPIRE titles and EPC export removes these approximations.

Fare zones are drawn from each station’s TfL zone, because TfL publishes no zone boundaries: every point takes the zone of its nearest station, and a boundary station counts as its lower zone. Treat the lines as a guide, and check a specific journey with TfL. Beyond Zone 6, zones are pockets around individual stations, not rings.

Posh Maps — the London property map. A demonstration build covering Greater London. Built with open data and MapLibre.