How Dwelfy Decides

What the chips mean

• Real (green) — pulled directly from a listing, the regulator, or HM Land Registry. We didn't compute it.
• Typical (amber) — market-average derived from comparable real listings; the method is described below.
• Estimate (rose) — a model output where we couldn't find enough comparables, or the comparable set was outside our standard radius. Verify before offering.

Rental estimates

estimated_rent is the conservative 25th-percentile floor of comparable rents in the same area, for the same property type and bedroom count. It is intentionally lower than what tenants typically pay so a yield that clears at this rent is a yield that clears.

1. Every active whole-property rental on Rightmove + Zoopla is aggregated into ~100 m spatial grid cells, stratified by (property_type, bedrooms). Houseshares are excluded; they feed a separate HMO grid.
2. Within each cell, we trim outliers to the 5th–95th percentile.
3. For each sale property, we run an expanding-radius spatial join — 0.5 km → 1.0 km → 1.5 km (hard cap). The smallest radius that passes all gates wins.
4. Gates: ≥ 2 qualifying cells, ≥ 6 underlying rental listings, and a homogeneity check (max cell-median ÷ min cell-median ≤ 2.5) so we don't mix a £900 cell with a £3,000 cell.
5. The figure shown is the 25th percentile of the cell medians that pass — a deliberately conservative floor.
6. When the gates fail we show “Not enough comparables” rather than guess.

rent_sample_size is the total number of real rental listings that backed the estimate. rent_search_radius_m is the winning radius (500 / 1000 / 1500 m). Both are visible next to the rent figure.

HMO rents

HMO per-room rents come from a separate SpareRoom + houseshare grid, processed the same way — stratified, trimmed, gated, conservative percentile. hmo_data_samples and hmo_data_radius_m are the equivalent provenance figures for the HMO estimate.

Because HMO and BTL rents are sourced from different markets, we never mix them. A flat that doesn't have enough houseshare comps within 1.5 km gets “not enough comparables” on the HMO line, even if BTL rent is well-evidenced.

Sold comparables

When we say “BMV −12% vs comps”, the comps come from HM Land Registry Price Paid Data — every residential sale registered in England & Wales since 1995. We compare against sales of the same property type with the same bedroom count within 0.5 mi over the last 36 months.

Contains HM Land Registry data © Crown copyright and database right (Open Government Licence v3.0).

Yields & cashflow

• Gross yield = annual rent ÷ price × 100. Uses estimated_rent, so a green / amber / rose chip on the yield matches the chip on the rent.
• Net yield = annual cashflow ÷ total cash in × 100. The full calculator deducts mortgage interest, letting-agent fee, void allowance, maintenance, monthly service charge / ground rent, 5% CapEx reserve and (HMO) full Council Tax / (BTL) void-prorated Council Tax.
• HMO net yield is the upstream-computed figure after conversion CapEx, running costs, voids and insurance — derived from the SpareRoom comp grid, not from the calculator's defaults.

Strategies in detail

Every property is modelled simultaneously across seven exit strategies. The headline yield is the best of them. Each strategy below explains the rent source, the cost model, the regulatory constraints, and when it applies.

BTL — standard Buy-to-Let

Single household on an Assured Shorthold Tenancy. Rent = local comparable median from estimated_rent (Q1–Q3 band shown alongside). Standard BTL mortgage products. Void allowance 2 weeks/yr default; agent management 10%; insurance + maintenance scaled by property value. Default for any property where a specialised strategy doesn't beat it.

LHA — Local Housing Allowance

Letting to housing-benefit tenants. Rent capped at the published LHA rate for the property's BRMA (Broad Rental Market Area) and bedroom count from the VOA annual table. Lower headline rent vs market (LHA is typically 30th-percentile of local market), offset by very low voids and council-direct rent payments via Alternative Payment Arrangement (APA). LHA cap is clamped to market rent when the BRMA cap is higher than local comps — so LHA-tenant rent is never artificially inflated.

Council — nomination lease

Direct lease (3–10 years) to the local authority. The council sub-lets to a household on its housing register, handles all tenant management + repairs + evictions, and pays guaranteed monthly rent regardless of occupancy. Rent = LHA × (1 + scheme premium %) where the premium is per-LA: 0% for Section-22 / PSL frameworks (Highland, Fife, Barnet), +5% (Brent), +10% (Barnsley, most Northern metros), +12% (Hackney), +15% (Westminster). Of 350 UK LAs covered: 210 run a direct scheme (council strategy enabled), 140 are verified to have no scheme (council strategy disabled and the tab is greyed out on the property page).

HMO — House in Multiple Occupation

Letting individual rooms to 3+ unrelated tenants from 2+ households. Higher gross yield vs BTL (typically 1.5–2.5×) but materially more management, regulation, and capex. Income split by room type using SpareRoom comp data: doubles, singles, ensuites, studios each carry their own rent. Single rooms typically rent 15–35% below doubles (up to −36% in Bath). Three regulatory tiers based on total rooms:

• 3–4 rooms (Small HMO, C4 use class) — permitted change of use from C3 unless property is in an Article 4 area; no mandatory licence (some councils run additional licensing schemes); standard fire safety.
• 5–6 rooms (Licensed HMO, C4 + mandatory licence) — HMO licence required (~£500–1,200 per 5y, varies by LA); L2 alarm system; annual gas safety, 5y EICR, PAT testing.
• 7+ rooms (Large HMO, Sui generis) — full planning permission required regardless of Article 4; specialist HMO mortgage only (most BTL lenders cap at 6 rooms); L1 alarm + often Sec 257 conversion rules apply.

Article 4 designations restrict permitted change of use from C3 to C4 in specific zones (universities, central London boroughs, several Northern metros). When a property is in an Article 4 area, the HMO yield is risk-weighted by the local council's historical refusal rate (or a conservative 50% fallback where decision data is unavailable). Properties NOT in Article 4 areas show the full unweighted HMO yield.

SA — Serviced Accommodation (short-let)

Furnished short-stay let via Airbnb / Booking.com. Highest gross yield potential but highest variance (ADR × occupancy × seasonality). Income = local SpareRoom-derived ADR × occupancy rate from comparable properties in the postcode sector; blended/proxy/baseline methodology depending on data density. Regulatory caps enforced server-side: London 90-day annual limit, Scotland short-term-let licensing, Wales statutory registration. VAT registration triggers at £90k/yr threshold (raised April 2024). Operating costs ~30% of gross (cleaning per stay, linen, OTA commission, utilities, mgmt fee).

BRRRR — Buy, Refurb, Rent, Refinance, Repeat

Buy below-market with bridging finance (typically 6–18 months, ~0.85% per month), refurb 3–9 months, refinance onto a long-term BTL mortgage at the post-refurb value — recycling deposit back out to fund the next deal. Post-refurb value derived from sold comparables in the postcode sector (HM Land Registry data); fallback to offer × 1.4 multiplier. Requires confident value-add plan and comparable evidence at refinance. Refinance LTV typically 75% (60–80% range investor-editable).

Flip — refurb-and-sell

Buy below-market, refurb 3–9 months, sell at uplifted value. No rental income. Exit price derived from sold comparables in the postcode sector + property type; fallback to offer × 1.25 multiplier. Carrying costs (insurance, utilities, council tax, finance) accrue during hold. Selling costs: 1.5–2% agent + £1,500 legal. CGT applies on the gain at 18% / 24% (April 2024 cut from 28%) depending on taxpayer band, plus the +5% additional-property SDLT surcharge on the purchase.

Each strategy's sub-tab in the property calculator lets you adjust every input and override per-line values. Hover any strategy tab on a property page for a pros / cons popover and the typical yield range for that strategy.

Updated 2026-05-27 (LEASE-CAL-1 real per-LA scheme premiums, SITE-77 HMO singles/doubles split, SITE-78 HMO planning regime tiers).

Deal score

Every property gets a 0–100 score, refreshed nightly. We score in the order property investors actually think about a deal: cashflow + yield first, capital growth second, risks deducted last.

Primary positives — cashflow + yield

• Gross yield — the dominant signal. Best of BTL, HMO, LHA, council-lease yields (roughly: 7% yield → 70 base).
• Monthly cashflow — the no.1 reason people invest. +£500/mo nets +10; +£200/mo nets +5. Any negative cashflow demotes the score with severity matched to magnitude: mild (under −£200/mo) deducts 10, significant (−£200 to −£500/mo) deducts 20, heavy (under −£500/mo) deducts 30 — the property would need monthly capital injection.

Secondary positive — capital growth

• 5-year IRR — total return on capital including exit. ≥10% nets +5, ≥6% nets +3. Negative IRR (capital outlay never recovered) deducts 30.
• Below-market-value — discount vs nearby sold comparables. ≥3% below vs 3+ comps nets +8.

Other positives

• Stress-test ICR pass at 6.5% (+ icon)
• Tenanted in situ — skips a void cycle (+4)
• Long lease (≥100yr) or freehold (+ icon)
• EPC A–C — MEES compliant (+ icon)
• Article 4 clear when HMO is feasible (+ icon)

Risk deductions

• EPC E (−10), F/G (−25)
• Lease tone “bad” (−15)
• Article 4 + HMO feasible (−6) — conversion blocked
• Auction lot (−5) — bridging-only finance, 28-day completion
• Cash-only listing (−5) — often signals cladding / short lease
• Fabricated rent (rent_sample_size = 0) (−10) — yield computed on a synthetic rent, not real comparables
• HMO net yield above the regional sanity ceiling (−5) — pipeline error or value-add audit needed

Hard caps

• Short lease (<80y on a leasehold) — score capped at 30 (mortgageability risk)
• EPC F or G — score capped at 35 (MEES blocker by 2030)
• Flood Zone 3, doubling ground rent, no-EWS1 mid-high-rise — filtered out entirely

Suspicious yields (above the per-region sanity ceiling — 18% in London, 22% Manchester / Leeds, 26% Liverpool / Newcastle) get partial credit only, so a single dodgy data point can't dominate. And we trust the tenure column over scraper lease fields — freehold properties never get phantom ground-rent / service-charge cost loaded onto them.

Updated 2026-05-22 (UI-7 cashflow penalty, UI-8 freehold guard, UI-9 cashflow + IRR positive bonuses).

SDLT / LBTT / LTT

Stamp duty uses the current statutory bands (England April 2025, Scotland LBTT, Wales LTT) plus the +5% additional-property surcharge and the +2% non-UK-resident surcharge. First-time-buyer relief is applied where eligible (we honour the FTB capture from onboarding).

Article 4 directions

We carry full PostGIS geometry for 4,313 designated Article 4 areas across the UK. If a property's lat/lng falls inside (or within 5 km of) a designated area for HMO conversion, the detail-page Article 4 card shows the actual designation name and distance.

Data sources & attribution

• HM Land Registry Price Paid — sold comparables. Open Government Licence v3.0; “Contains HM Land Registry data © Crown copyright and database right”.
• EPC Register (DLUHC / gov.uk Open Data Communities) — every EPC certificate lodged since 2008. OGL v3.0.
• Rightmove + Zoopla + OnTheMarket — listings indexed from public listings; we add analytics and don't republish theirs.
• SpareRoom — HMO room rents.
• Ofsted (GIAS) — schools. OGL v3.0.
• Environment Agency — flood zones. OGL v3.0.
• police.uk — area crime stats. OGL v3.0.
• VOA — council tax bands.

Freshness

Every property row carries a roi_calculated_at timestamp — visible on the detail page as “Refreshed Xd ago”. The full ROI bundle (rent estimate, HMO numbers, yield, deal score) is rebuilt nightly from the latest scrape. If you see a stamp older than 7 days, treat the numbers with caution and re-scan.

What Dwelfy is not

• Not FCA-regulated. Research tool, not financial advice. Numbers are point-in-time estimates that you should verify before exchanging contracts.
• Not an AVM for individual properties. We don't compute “this exact flat will let for £X” — we compute “a reasonable rent for a 2-bed flat in this 100–1500 m area is at least £X”. For a unit-specific valuation, instruct a RICS surveyor.
• Not a substitute for a solicitor, building survey, EICR, gas safety check, EPC review, or planning consultation. Especially on leasehold flats and HMO conversion plays.

Push back

If you think a number is wrong, tap the chip next to it, look at the comp count, the radius, and the freshness stamp, then tell us at hello@dwelfy.co.uk. Every report goes to the dataset owner.

Full audit findings: Glossary