How it works
How the price check works
A transparent, reproducible price check built only on registered transactions. Buyers and sellers get the same numbers, and we say how much evidence there is.
Jump to the methodology summary
Step 1
Official sales only
Registered HDB and URA sale prices only — never asking prices.
We use registered HDB resale transactions from data.gov.sg and, for private homes, transaction records from URA. Each record has a month, a floor range (not the exact floor), the floor area and the price — but no unit number, facing, view or condition. Asking prices are never used as evidence. Every report says how recent its data is.
Step 2
Closest sales first
Same block or project first; we widen the search step by step only when there are too few sales.
We start with the sales most like this home: the same block for an HDB flat, the same project for a condo. Only when there are too few do we widen the search, step by step — neighbouring blocks or projects, then the wider neighbourhood, then the whole town or postal district. The full report shows how many sales came from each step and why the search widened. We use resales only: developer (new-launch) sales are shown as background, never in the price, and sub-sales are left out. For private homes, the same tenure comes first; executive condos are compared only with other ECs, at the same stage first.
Step 3
Scoring similarity
Up to 20 sales, scored on distance, size, lease, floor and recency. A range needs at least 3.
Each sale gets a similarity score from how close it is, its size, its lease (or tenure), its floor and how recently it sold. We keep up to 20 of the closest, and show a range only when at least 3 similar sales support it.
Step 4
Adjusting for market movement
Older sales are brought to the latest official price index quarter. No forecasting.
Prices move over time, so each older sale is adjusted with an official quarterly price index for that kind of home before it is compared. Official indices come out with a delay, so sales are brought to the latest published index quarter — usually one or two quarters before the month of our data (HDB flats are carried one quarter further with the index's latest change). Every report names that quarter. Sales since then are not adjusted, so the reference price reflects the market as of that quarter plus the recent sales themselves — it is not a forecast. In a rising market, later sales will tend to be higher.
Step 5
The reference price and range
The mid reference price is the weighted middle of adjusted sales; the range is as wide as past accuracy needs.
The mid reference price is the similarity-weighted middle of the adjusted sales. The range is checked against how past sales of similar homes actually turned out, so it is as wide as the evidence needs — no wider to look safe, no narrower to look precise. When there isn't enough evidence, we say so instead of showing a range.
Step 6
Evidence and accuracy
Two answers instead of one score: how strong the evidence is, and how the method did before.
Instead of one “confidence” score, each report answers two questions. Evidence: how many close, recent sales there are, how much they tell us together (20 sales may be about as informative as 15 equally similar ones), and how much of your home’s details we have. Track record: how our method did on the past sales the range was set from, e.g. “On the past sales this range was set from, 8 in 10 sold within ±6% of our mid reference price”, and separately how it did when we tested it on newer sales.
Step 7
Evidence for your offer
Where the asking price sits among the comparable sales. We never suggest an offer price.
The full report shows where the asking price sits among the comparable sales — how many sold for less and how many for more — and lists them. We do not suggest an offer price: past sales are evidence for your negotiation, not a target.
Step 8
When we show comparable sales only
Landed homes, projects with no resales yet and ECs that just changed stage get comparable sales, not a range.
For terrace, semi-detached, detached and strata landed homes, official records mostly give the land area, not the built-up area. A price per square foot can't tell a bigger house from a bigger plot, so these reports show the closest sales instead of a range. A project with no resales yet, or an EC that just changed stage, gets the closest resales of the same kind — condos for a condo, ECs for an EC. Developer sale prices appear as background only.
Step 9
What we leave to you
Facing, view and renovation are yours to judge at the viewing. We report past sales; we don’t forecast prices.
Official records don't capture facing, view or renovation, so we don't turn them into dollar adjustments — you judge them at the viewing. We report what past sales support; we don't forecast prices.
Step 10
Reproducible
Every check keeps its inputs, the exact sales used and the data version, so it can be traced and re-run.
Every check stores its inputs, the exact comparable sales, the data and index versions, and the method and calibration versions, so it can be audited and reproduced. Each report ends with its data cut-off, the number of sales used and its sources.
We re-run the method at past dates using only the data available then, and compare it with what homes of each type later sold for.
Accuracy
Within 3.3%Half of later HDB resales sold within 3.3% of our mid reference price.
Apr–Sep 2026 · 13,276 later sales · HDB resale
Like an exam with the answers covered: each sale was priced using only data from before it, then compared with what it actually sold for.
- Within ±3.3%1 in 2 (50 of 100)
- Within ±5%2 in 3 (67 of 100)
- Within ±10%9 in 10 (91 of 100)
Real sales
We priced them first. Then they sold.
HDB · 5-room flat · Bedok
- Our mid price
- S$656k
- Sold Sep 2026
- S$635k
3.3% apart
Condo · Stirling Residences
- Our mid price
- S$1.59M
- Sold Jul 2026
- S$1.54M
3.2% apart
Real past sales, not customer stories — one typical sale per home type, not our best. Across all later sales, half of HDB resales sold within 3.3% of our mid price and half of condo resales within 3.2%.
See how accurate we areMethodology summary
The key rules in one place, applied the same way to every check.
- Data
- HDB resale transactions (data.gov.sg) and URA private residential transactions. Sale prices only, never asking prices.
- Resales only
- The price check uses resales. Developer (new-launch) sales are shown as background only; sub-sales are left out.
- Comparable sales
- Same type of home, closest first (same block or project), widening only when there are too few. Last 24 months, up to 20 sales.
- Executive condos
- Compared with other ECs at the same stage (minimum occupation period, past MOP, fully private), never with private condos.
- Similarity
- Location, size, lease or tenure, floor and how recently it sold.
- Time adjustment
- Each sale brought to the latest published quarter of an official price index. The report names the index and any quarter extended. No forecasting.
- Reference price
- Similarity-weighted middle of the adjusted sales, times your home’s floor area.
- Reference range
- Calibrated so that about 8 in 10 sales of similar homes in the build period landed inside. Shown only when enough similar sales support one. On later sales (Apr–Sep 2026): 76 in 100 HDB resales and 79 in 100 condo resales landed inside.
- Conclusion
- One sentence (Below range, Fair, Slightly high or High) and up to 3 reasons, each from the numbers by fixed rules.
- Observed differences
- How nearby sales differ by floor, lease or size. A description, not cause and effect, and not used for the price.
- Evidence
- How many close, recent sales there are, how many come from the same block or project, and how complete your details are.
- Accuracy
- Measured on past sales of the same kind of home, using only the data available at each past date.
- Landed homes
- Comparable sales only — no reference price or range.
- Offer evidence
- Where the asking price (or a seller’s target price) sits among the comparable sales. No suggested offer or listing price.
- Buying or selling
- Same numbers either way. Buyers see whether the asking price fits recent sales; sellers see what recent sales support.
- Condition
- Optional, chosen in the report. It never moves the range — add your own renovation budget to compare total cost.
- Sharing
- Your report page opens only in your browser. Share it by link: people see the quick price check, never your number. Turn it off any time.
Sources and licences
- Contains information from the “Resale flat prices” dataset from data.gov.sg, which is made available under the terms of the Singapore Open Data Licence version 1.0. Read the licence
- Private residential transactions: Urban Redevelopment Authority (URA) Data Service.
- Contains information from the “HDB Resale Price Index” dataset from data.gov.sg, which is made available under the terms of the Singapore Open Data Licence version 1.0. Read the licence
- Contains information from the URA private residential property price index datasets from data.gov.sg, which are made available under the terms of the Singapore Open Data Licence version 1.0. Read the licence
- Locations: OneMap, Singapore Land Authority.
- SG Home Price Checker is independent. It is not affiliated with, and is not endorsed by, HDB, URA, SLA or any other Singapore Government agency.