HDB flat offer too low? 5.2% of matched four-room resales went 10% or more under

A proposed price is too low when it sits well under recent resales of flats that truly match yours: the same town, a lease that began within the same five years, a similar storey and a similar floor area. Among four-room resales matched on town, lease band and storey, about one in 20 went 10% or more below its group's median 1.

That is the short answer for a household that has just been offered a price for its flat. The rest of this guide sets out a method built only from official data. It covers how to choose comparable sales, how wide their range normally is, which differences move the price, what HDB's valuation does for a seller, and how the test works for a condo. It does not say whether any offer should be accepted.

What counts as a comparable sale for an HDB flat?

HDB publishes its resale flat transactions on data.gov.sg, by the month the resale was registered, with the town, flat type, block, street, storey range, floor area in square metres, flat model and the year the lease began 1. Those fields are the building blocks of a fair comparison. HDB's notes on the dataset add two cautions. The records leave out resales that may not reflect the full market price, such as sales between relatives and of part shares. And the prices "should be taken as indicative only as the resale prices agreed between buyers and sellers are dependent on many factors" 1.

A comparable sale is one that differs from your flat in as few of those fields as possible. In practice that means four filters, applied in this order:

1. Same flat type in the same town. A four-room flat in one town is not evidence about a four-room flat in another.
2. A lease that began in the same five-year band. The worked example below shows how far the lease band alone can move a median 1.
3. A similar storey range. HDB reports storeys in bands of three, such as 07 to 09 1.
4. A similar floor area. Within one flat type, floor areas can differ by 20 square metres or more, as the worked example below shows 1.

Then there is time. A registered resale shows what was paid, but the price is agreed earlier, in the Option to Purchase 4, so even the newest record reflects a price agreed before it was registered 1. Older records need adjusting for how the market has moved since, which is covered further down.

The catch is supply. Of the 10,811 four-room resales registered from October 2025 to September 2026, 19.0% had no other four-room resale in the same block over those 12 months 1. The median sale had two others in its block 1. Stretching the look-back to 24 months, from October 2024, raises the median to four other sales in the block and cuts the share with none to 7.0% 1. Most sellers will need to look beyond their own block, so the question is how far to widen.

How much do close comparable sales vary?

Less than most sellers expect, once the comparison is tight. Sightlines grouped all four-room resales registered from October 2025 to September 2026 in six different ways, from the loosest (same town) to the tightest (same town, lease band and storey range). For each group with at least 10 sales, it measured the spread of the middle half of prices, from the lowest-quarter mark to the highest-quarter mark, as a share of the group's median. It then took the median spread across groups, and counted how many sales in those groups went 5% or 10% below their own group's median 1.

Source: HDB resale flat prices, data.gov.sg 1; Sightlines analysis. Lease bands are five calendar years by the year the lease began (for example 1985 to 1989). Spread is the gap between the 25th and 75th percentile prices divided by the median, taken as the median across groups.

Two lessons come out of the table. First, a town-wide median is a poor yardstick. Across a whole town, the middle half of four-room prices spans 17.8% of the median, and one sale in five went 10% or more below it 1. That is mostly because older and newer flats, and low and high floors, are mixed together.

Second, once lease and storey are matched, the middle half narrows to about 7% of the median 1. In the two sets that match the lease band as well as the storey range or the street, 5.2% and 5.7% of sales went 10% or more below their group's median, roughly one sale in 20 1. Going 5% under was far more common: 18.8% and 19.8% of sales did, roughly one in five 1.

That gives a seller a scale. A price about 5% under a tight comparable median is within the ordinary range of registered resales 1. A price 10% or more under it is unusual, and worth asking about before an Option to Purchase is granted 1. Neither is proof on its own: the flat's condition, its facing and the urgency of either side are not in the records.

How much do storey, lease and size move the price?

Storey. Within each town and lease band, Sightlines compared the median four-room price in each storey range with the median for storeys 04 to 06. It used every group with at least five sales in both ranges and took the median ratio across groups 1.

Source: HDB resale flat prices, data.gov.sg 1; Sightlines analysis. Groups are the same town and five-year lease band. Higher storey ranges had fewer than 10 groups to compare and are left out.

A seller on the 11th storey who is shown only low-floor sales may see an offer that looks fair against them but sits a few per cent under what higher floors fetched.

Lease and location within a town. In Tampines, four-room flats whose leases began from 1985 to 1989 had a median of $618,000 across 258 resales registered from October 2025 to September 2026 1. Those whose leases began from 2020 to 2024 had a median of $770,444 across 250 resales, 24.7% higher 1. Lease is not the whole story, though. The 22 resales of flats with leases from 2010 to 2014 had a higher median still, $861,444, than the newest band 1. Location within the town and the design of a project can matter as much as the lease, which is why the same street is a useful filter when there are enough sales on it.

Floor area. Within the same Tampines lease band, the 103 resales at 104 square metres had a median of $643,000, and the 65 at 84 square metres had a median of $560,000 1. The larger flats sold for 14.8% more 1. Both are labelled four-room in HDB's records. A seller with a 104-square-metre flat who is shown 84-square-metre sales as comparables is being shown the wrong flats.

A worked example: one four-room flat in Tampines

Take a household selling a four-room flat in Tampines on a lease that began in the late 1980s, on the eighth storey (so storeys 07 to 12, the bands either side, are used), with a floor area of about 104 square metres. Each step below adds one filter to the four-room resales registered in Tampines from October 2025 to September 2026 1.

Source: HDB resale flat prices, data.gov.sg 1; Sightlines analysis. The lowest-quarter and highest-quarter marks are the 25th and 75th percentile prices.

All 71 sales in the last row are the Model A layout, and 47 of them are exactly 104 square metres 1. That matches every field the records hold.

Now suppose a buyer proposes $620,000, the lowest-quarter mark for all Tampines four-room resales 1. Against the town-wide median of $665,000, it looks 6.8% low 1. Against the median for its lease band, $618,000, it looks slightly generous, at 0.3% above 1. Against the 71 closest comparables, it is 4.6% below their median of $650,000 and $15,000 under their lowest-quarter mark of $635,000 1. Fourteen of those 71 sales, or 19.7%, were registered at $620,000 or less 1.

So the same proposed price can read as low, fair or generous depending on the comparison. On the tightest set it sits in the bottom fifth of recent sales of near-identical flats. That is not proof that it is too low. It is a reason to ask what the buyer has seen that the records do not show, such as a dated kitchen or a long extension of stay. The same check takes minutes for any flat in the dataset.

Does HDB's valuation protect a seller?

Not in the way many sellers assume. Under HDB's terms and conditions for resale flats, the price is fixed before any official value exists.

• The seller and buyer must use HDB's prescribed Option to Purchase. The Option Fee may not exceed $1,000 and the deposit may not exceed $5,000, and the Option Period is 21 calendar days after the Option Date 4.
• Only a buyer who has already been granted an Option to Purchase may submit a Request for Value, and must do so by the next working day after the Option Date 4.
• HDB decides whether a valuation is needed, and if so assigns a valuer from its panel. The seller must let the valuer in within three working days of being told 4.
• The Value "will form the basis for his CPF usage and the reference for the housing loan amount", in the words of the terms. It applies to a buyer financing with CPF savings or a housing loan 4.
• The seller and buyer "must declare to HDB the true resale price of the flat", and any side arrangement that inflates or understates it is not allowed 4. If the declared price is below the Value, HDB may ask for a written explanation, and may withhold acceptance of the resale application or cancel it if the reason is not acceptable 4.

The order matters. The Value arrives only after the seller has granted the option, so by then the price in the option cannot change: HDB's terms say there shall not be any change to the declared resale price 4. The Value exists to set the buyer's CPF usage and loan, not to tell a seller whether the price was fair. The comparable-sales check above is the seller's own test, and it has to be done before the option is signed.

How should older comparable sales be adjusted?

By how much the market has moved since they were registered, using HDB's resale price index. The index tracks overall price movement across towns, flat types and models, and HDB explains it as a ratio: if the index rises from 100 to 108, prices on the whole rose by 8% 3.

The adjustment is a single multiplication: the comparable price, times the latest index, divided by the index for the quarter in which the comparable was registered. HDB's flash estimate for the third quarter of 2026 is 202.4, down 0.2% on the quarter 3. The fourth quarter of 2025 stood at 203.6 2. A $650,000 resale registered in late 2025 therefore adjusts to about $646,169 at the third-quarter level, a fall of 0.59% 23. The third-quarter figure is a flash estimate 3.

Today that correction is small. The index fell by 0.1% in the first quarter of 2026 and by 0.3% in the second 3. In a fast market it is not small. From the third quarter of 2021 to the third quarter of 2022, the index rose from 150.6 to 168.1, or 11.6% 2. A seller then relying on year-old comparables without adjusting would have been measuring offers against prices more than a tenth below the market of the day. The rule is simple: the older the comparables, the more the index matters, and the more the market is moving, the shorter the look-back should be.

The index is a whole-market average, so it corrects for timing; it is not a substitute for recent sales.

Does the same method work for a condo?

Yes, with one change: compare price per square foot inside the same project, and keep the quantum beside it. For private homes, URA's private residential transactions record each sale's contract date, project, sale type, area and price 5. A new sale is a unit bought from the developer; a sub sale is a unit resold before the project's completion; a resale is a unit resold after it. Quantum is the total price paid. Psf is that price divided by the strata area URA records, converted to square feet 5.

Sightlines ran the same narrowing on condo and apartment resales contract dated October 2025 to September 2026 5:

Source: URA Data Service, private residential transactions 5; Sightlines analysis. Riverfront Residences is a 99-year project in District 19 whose lease began in 2018, and had the most resales of any project in District 19 over the window 5. It is used only to show the method.

Inside one project, the middle half of resale prices spanned only 8.0% of the median psf, against 28.4% across the whole Outside Central Region 5. Quantum is a different matter. The project's lowest-quarter and highest-quarter prices were $860,000 and about $1.61m, because the units range from small to large 5. A condo seller should therefore check the offer's psf against same-project resales of a similar size, and then check that the quantum makes sense for that size. Psf alone can flatter a small unit and penalise a large one.

Two cautions apply. Floor, facing and stack differ within a project in ways the aggregates do not show. And the latest month is still being reported, so URA's newest figures are provisional and are left out here 5.

What would change the answer?

A comparable-sales check tells a seller where an offer sits among recent sales. It does not set the flat's value, and three things can break it.

The flat is not like its comparables. A renovated flat, an unblocked view or a corner unit may justify a price above the tight median. A flat that needs work, faces a busy road or comes with a long extension of stay may not reach it. The records carry none of this 1. Adjust the reading, and say why, before treating a gap as a lowball.

There are too few comparables. A group of fewer than 10 sales gives a median but no reliable spread, and a median of a handful of sales can be one unusual flat. Widen the time window first and adjust it with the index, then widen to the same street, and only then to the town with lease and storey matched.

The market is turning. HDB's index fell slightly in each of the first three quarters of 2026, by the flash estimate for the third 3. In a falling market, unadjusted comparables from a year ago will make every offer look low. In a rising one, the opposite.

A practical sequence for a seller weighing an offer:

1. Pull the registered resales for your flat type and town from the last 12 months on data.gov.sg 1.
2. Filter to your five-year lease band, then to your storey range or the ranges either side, then to floor areas within about five square metres of yours.
3. Stop narrowing while at least 10 sales remain. Note the median and the lowest-quarter mark.
4. Adjust older sales with HDB's resale price index 23.
5. Read the offer against that range. About 5% under the median is ordinary, and about 10% under is unusual, worth questioning before an Option to Purchase is granted 1.

Turning down an offer is also a choice with a cost: the flat stays on the market, and the next offer could be higher or lower. The check does not decide that trade-off. It puts the right numbers in view.

This is general information, not financial advice. Check the current rules with IRAS, HDB, CPF Board or your bank before you commit.