By Gary Lim · ERA Senior Division Director · CEA R009877B · Updated 19 September 2026

The thing nobody will admit first

Every guide on page one of Google answers how long to sell an HDB flat with the same number: eight to twelve weeks. That number is real. It is also an answer to a completely different question.

Eight to twelve weeks is the administrative timeline. It starts on the day a buyer signs your Option to Purchase. It is HDB paperwork, a portal submission, a completion appointment. It is predictable, it is bounded, and it is the easy half.

The hard half — the half you actually lie awake about — is the stretch between the day your listing goes live and the day someone hands you an option fee. Nobody on page one puts a number on that, because nobody can. Here is why.

Singapore’s public resale registry records the sale. It does not record the listing. There is no public days-on-market series for HDB the way there is for, say, US residential. HDB does not publish it. The portals do not publish it in any auditable form. So when a competitor writes that a flat takes “two weeks to three months” to find a buyer, that is not a measurement. It is a shrug with a range attached.

I am not going to shrug. I am going to tell you what cannot be known, then give you the honest substitute: the base rate. If I cannot tell you how long your specific flat will take, I can tell you exactly how big the queue is, how fast the queue moves, and roughly where in it you are standing. That is a real answer, and as far as I can find, nobody else has published it.

Everything below comes from 240,807 HDB resale registrations on the public registry — every one recorded from January 2017 to the part-month of September 2026 — plus HDB’s own dwelling-units-by-town table. I pulled it all on 19 September 2026. I read every published integer at least twice, and I show my reconciliation checks so you can catch me if I am wrong.

At a glance

The base rate: 1 flat in 42

Start with the denominator, because it is the part everyone skips.

HDB’s dwelling-units-by-town table, filtered to sold units only, gives 1,032,699 flats across the 26 towns and estates in the latest published financial year. That is the pool. Every one of those households could, in principle, decide to sell tomorrow.

Now the numerator. I counted the registry month by month for the twelve months from September 2025 to August 2026, the last complete month available:

24,590 sales against 1,032,699 flats is 2.38% a year. One flat in forty-two. Roughly 2,049 resales register in an average month across the whole island.

Translate 1 in 42 into something you can picture. Stand in front of a 100-unit block. Over the next twelve months, between two and three of those front doors will change owner. Not twenty. Two or three.

That single fact reframes the whole question. Selling an HDB flat is not like selling a used car into a thick daily market. It is a slow, thin market where your unit competes for a trickle of buyers who are looking at that block, that flat type, that price band, in that particular month.

How long to sell an HDB flat by flat type

Here is where I think this piece earns its keep. When I measured turnover by town, I found the towns collapse into a narrow band. But you are not a town. You are a flat type. A 3-room seller in Toa Payoh and an executive seller in Pasir Ris are standing in two completely different queues, and I could not find anyone who had measured them separately.

So I did. Twelve-month transaction counts by flat type, divided by the stock of each type from the dwelling-units table:

Look at the spread. From the slowest type to the fastest, the whole range is 0.24 percentage points. Executive 2.26%, 4-room 2.50%. One in forty-four against one in forty.

If you are an executive-flat owner who has been told your flat is hard to shift, that is the honest scale of the handicap: about nine percent fewer sales per hundred units per year than a 4-room. It is real. It is not the cliff you were promised.

I was wrong about executives

I want to be straight about this, because I went into the analysis with a firm view and the data only half-agreed with me.

My prior, from seventeen years of doing this, was that executives and 5-rooms turn over much more slowly than 4-rooms. Bigger quantum, thinner buyer pool, more CPF and loan friction, fewer families that need 130 square metres. I expected a gap of something like a full percentage point.

I got 0.24 points. The direction of my prior was right — executive is genuinely the slowest of the main types, 5-room is second slowest, 4-room is fastest — but the magnitude was wrong by a factor of four.

Why was I wrong? Two reasons I can defend. First, the stock is small: there are only 65,116 executive flats in the country against 433,178 4-rooms. A thin buyer pool chasing a thin stock does not have to produce a low turnover rate. Second, executive owners are disproportionately at the life stage where they sell — older households downsizing — which pushes the numerator up even as the buyer pool stays narrow. Those two effects partly cancel.

The practical lesson for an executive or maisonette seller: your problem is almost never “nobody wants this size”. Your problem is price discovery, because there are so few comparable transactions that neither you nor the buyer has much to anchor on. I go into that in more detail in my piece on HDB executive maisonette prices in 2026.

Flat type inside five towns

The national table hides something. Turnover by flat type is nearly flat nationally, but that is an average of averages. Inside a single town the picture changes completely.

I built the cross-tab for five towns chosen to make the point: the fastest-turnover town and the slowest from my earlier town analysis, a big old town, a big young town, and Pasir Ris, which has an unusual stock with roughly a quarter of its sold units being executives.

Three things jump out.

The cross-tab band is far wider than either margin. Nationally, flat type spans 0.24 points and my five towns span 1.18 points. But the individual cells run from 1.40% (Queenstown executives) to 4.29% (Pasir Ris 3-rooms) — nearly three times. Neither “which town” nor “which flat type” tells you much on its own. The combination tells you a lot.

My prior holds in some towns and inverts in others. In Punggol, executives turn over at 2.04% against 3.46% for 4-rooms — a gap of 1.4 points, exactly the pattern I expected. In Ang Mo Kio, 5-rooms (2.20%) are the fastest type in the town and 3-rooms (1.77%) the slowest. In Pasir Ris the executives (2.43%) beat the 5-rooms (2.41%). The national flatness is not because flat type does not matter. It is because flat type matters in opposite directions in different towns and the effects cancel.

Some cells are too thin to trust. Queenstown recorded five executive sales in a year, from a stock of 356. Ang Mo Kio recorded nine. Pasir Ris recorded 21 three-room sales from a stock of 490. Treat those percentages as indicative, not precise — one extra deal moves a 5-sale cell by twenty percent. I have left them in because hiding thin cells is how you end up believing your own noise.

If your town is not in the table, the town-level numbers for several more are in my write-ups on Bedok HDB resale and Hougang HDB resale, and the full 26-town turnover series sits in my HDB resale volume by town analysis.

Months of inventory: your actual queue

Turnover tells you how much demand exists. It does not tell you how long you will wait, because it says nothing about how many other sellers you are queuing behind.

For that you need the listing side. Listings are not published as official statistics, so this is the one place in the article where I am leaning on a commercial portal. As at 19 September 2026, PropertyGuru’s own HDB-for-sale page headed itself with 13,653 flats listed. I read it twice and got the same number both times. Caveats, plainly: it is one portal, not the whole market; it will include some duplicate and stale listings; and it moves every single day. Treat it as an order-of-magnitude input, not a precise one.

Against 2,049 sales a month, 13,653 listings is about 6.7 months of inventory. In plain English: if every seller currently listed stayed put and no new seller joined, it would take roughly six and a half months to clear the board at the current rate of buying.

That is the single most honest one-number answer to how long to sell an HDB flat that I can construct from public data. And it immediately tells you something the eight-to-twelve-week crowd never will: the median seller is not waiting two weeks. The median seller is in a queue measured in months.

Three worked examples, every step shown

The arithmetic is simple enough to do on your phone. Four steps.

1. Find the stock of your flat type in your town (dwelling-units table).
2. Find the twelve-month sales of your flat type in your town (registry).
3. Divide sales by twelve to get monthly flow.
4. Divide the current listing count for your type and town by that monthly flow. That is your months of inventory.

Example 1 — a 4-room in Bedok

Stock: 21,971 four-room flats. Sales in the twelve months to August 2026: 496. Annual turnover 2.26%, or 1 in 44. Monthly flow: 496 ÷ 12 = 41.3 sales a month. PropertyGuru showed 208 four-room Bedok listings on 19 September 2026 (read twice, 208 both times). Months of inventory: 208 ÷ 41.3 = 5.0 months.

Example 2 — a 4-room in Punggol

Stock: 24,632. Sales: 852. Turnover 3.46% — the fastest cell of any I measured among the big types. Monthly flow: 852 ÷ 12 = 71.0 sales a month. Listings: 465 (two of three reads gave 465, one gave 473; the count drifts intraday). Months of inventory: 465 ÷ 71.0 = 6.5 months.

Stop on that for a second. Punggol 4-rooms turn over 53% faster than Bedok 4-rooms. And the Punggol seller is in a longer queue. Demand is higher, but supply is higher still, because Punggol is deep in its post-MOP wave and thousands of owners hit the five-year mark at once. High turnover is not the same as a fast sale. If you are in that wave, my guide to selling an HDB flat after MOP walks through the sequencing.

Example 3 — a 3-room in Ang Mo Kio

Stock: 24,443. Sales: 433. Turnover 1.77% — the slowest cell among the big types in my five towns. Monthly flow: 433 ÷ 12 = 36.1 sales a month. Listings: 290 (reads of 290 and 291). Months of inventory: 290 ÷ 36.1 = 8.0 months.

Eight months of inventory. That is the queue an Ang Mo Kio 3-room seller is actually joining, and no amount of staging changes it. What changes is where in the queue you stand, which is entirely a function of price relative to the other 289 flats.

Is there a best month to list?

Every seller asks. Almost every article answers with vibes. So I pulled national registration counts for all nine complete calendar years in the registry, 2017 through 2025 — 222,067 transactions — and computed each calendar month’s share of annual volume.

So: there is a real seasonal pattern, and it is smaller than the advice industry pretends.

February is genuinely quiet. At 6.91% of annual volume, index 83, it is 17% below an even month and the only month that stands clearly apart. Chinese New Year does what you would expect: viewings stop, offices close, families have other things on.

July and August are genuinely busy. Index 117 and 112. Together those two months carry 19% of the year against the 16.7% an even split would give them.

The other nine months are noise. January, March, June, September, October, November and December all sit between index 98 and 105. April and May dip slightly to 94 and 93. If someone tells you October is a bad month to list, they are describing a two-percent effect and charging you for the insight.

Peak to trough, the whole swing is a factor of 1.41. And that overstates the decision you actually face, because you do not sell into a single month — you sell into the three or four months after you list. Smooth it into a rolling three-month window and the best listing month (June, index 111) beats the worst (January, index 92) by about twenty percent of relative buyer flow.

My verdict on timing: ignore it. Twenty percent more buyer flow sounds like a lot until you set it against a pricing error. Overprice by three percent in a thin market and you will lose more than twenty percent of your buyer pool on the first weekend. Do not hold a flat off the market from January to June to chase an index of 111. List when you are ready, priced correctly, and you will beat every seasonal effect in this table.

What the February dip really means

One caveat that matters, and that I have not seen anyone else state when they wave at HDB seasonality.

The month in the registry is the month the resale registration lands, not the month the deal was struck. Between a buyer signing your Option to Purchase and the transaction appearing in the data, there is the option period, the resale application, and HDB’s processing. Realistically the registry month sits one to two months after the handshake.

So the February trough is not telling you that February is a bad month to list. It is telling you that December and January are slow months to find a buyer in — the pre-CNY lull plus the school holidays plus everybody being away. Likewise the July and August peak reflects deals agreed around May and June.

Shift the whole table back by roughly six weeks in your head and the actionable version is: the months in which buyers are most active are around May and June; the months in which they are least active are around December and January. Which, conveniently, matches what the rolling three-month calculation above gave me independently. Two routes to the same answer is worth something.

I wrote a longer piece on listing timing at /best-time-to-sell-hdb/ — that URL is not live yet, so do not go looking for it; when it publishes it will carry the town-by-town version of this table.

The administrative half: 21 days, 28 days, 8 weeks

Now the part everyone else leads with. It deserves proper treatment — just not first billing.

Once a buyer commits, here is the sequence, with HDB’s own stated timings:

• Option to Purchase granted. You grant the OTP and collect the option fee. HDB states plainly: the OTP is valid for 21 calendar days from the date it is granted. The buyer exercises within that window or the option lapses.
• Exercise. The buyer signs and pays the option exercise fee. Option fee plus exercise fee are capped by HDB — check the current ceiling on the HDB page, because it is the kind of number that changes without fanfare.
• Resale application. Both sides submit their portion through the HDB resale portal. Each side pays an administrative fee: $40 for 1- and 2-room flats, $80 for 3-room and bigger, inclusive of GST and non-refundable. Yes, both of you pay. No, it is not refunded if the deal collapses.
• HDB acceptance. HDB aims to notify both parties that the application has been accepted within 28 working days. In practice it is usually much faster, but 28 working days is roughly five and a half calendar weeks and you should plan for it rather than be surprised by it.
• Completion. HDB states that resale completion takes about 8 weeks after the application is accepted. Keys change hands at the completion appointment.

Add it up honestly. Three weeks of option period, up to five and a half weeks to acceptance, then eight weeks to completion. That is a worst-case administrative path of around sixteen to seventeen weeks, not eight to twelve. The typical path is shorter because acceptance usually comes well inside 28 working days, but if a competitor’s article told you eight weeks from OTP to keys and you have already committed to a new lease, you now know where the risk sits.

The full stacked timeline

Put both halves side by side. This is the table I wish somebody had shown me when I started.

The administrative half is three to four months and it is almost entirely out of your hands. The marketing half is anything from a week to a year and it is almost entirely in your hands. Every hour you spend worrying about the paperwork is an hour not spent on the only variable that matters.

What actually changes how long to sell an HDB flat

Seventeen years, five hundred-odd transactions. Here is my ranked list of what actually moves the needle, with the data reasoning attached where I have it.

1. Price against recent same-block transactions, not against asking prices

This is ninety percent of it. The mistake I see every week: a seller surveys the live listings in their block, finds the highest asking price, and anchors there. But asking prices are a list of people who have not sold. Transacted prices are a list of people who have. Price against the second list.

And remember the base rate. Your block generates two or three transactions a year. If the most recent one was eight months ago and slightly odd — a high floor, a renovated unit, a rushed sale — that single data point is doing all the anchoring work for every buyer who walks through your door. Know what it is before they tell you. My running commentary on where prices actually are sits in HDB resale prices 2026.

2. Whether your block has any recent price evidence at all

This is the underrated one and it falls straight out of the 1-in-42 arithmetic. In a 120-unit Ang Mo Kio 3-room block at 1.77% turnover, you can expect about 2.1 sales a year. In a 120-unit Punggol 4-room stack at 3.46%, about 4.2. The Punggol buyer has roughly twice as much evidence to anchor on.

When there is no evidence, buyers do not guess upward. They discount for uncertainty, or they walk to a block where they can see the numbers. If your block has been silent for a year, you are not just selling a flat, you are establishing a price. Budget time for that.

3. Floor and facing

Still real, still underpriced by sellers on low floors and overpriced by sellers on high ones. A high floor with an unblocked view in a dense estate can carry a genuine premium. A second floor facing a bin centre cannot be staged out of existence. Price it in from day one rather than discovering it across three months of dead viewings.

4. Realistic valuation versus COV fantasy

Cash over valuation is a negotiation, not an entitlement. The HDB valuation is requested after the OTP is granted, so both of you are committing before you know the number. A seller who has built their whole plan around a large COV is a seller who will be renegotiating in week four. Run your own proceeds maths before you list — the calculators at listings.sg/tools will do it in a couple of minutes, including the seller’s stamp duty exposure if you bought recently, which you can read about in seller stamp duty Singapore.

5. Agent activity in the first fourteen days

Listings get their attention in the first two weeks. A flat that goes live with poor photos, no floor plan, and a price that needs a correction burns that window and then competes as a stale listing for months. You do not get the first fortnight back.

Lease decay kills your youngest buyers first

This one is counter-intuitive enough that it deserves its own section, and it is a finding I have not seen anywhere else stated this way.

Everyone knows a shorter remaining lease narrows your buyer pool. Almost everyone gets the direction wrong. The intuition is that old flats become the preserve of the young, who cannot afford anything else. The mechanics say the opposite.

CPF usage for a resale flat is governed by whether the remaining lease covers the youngest buyer to the age of 95. A flat with 60 years left covers a 40-year-old to 100 — fine. It covers a 28-year-old only to 88 — not fine, and their CPF usage gets restricted, which means more cash up front, which for most young couples means they cannot buy it at all.

So lease decay strips out your youngest buyers first, and works upward. The buyers you lose are precisely the ones with the longest earning runway and the biggest CPF balances relative to their needs. By the time a flat is under about 60 years, you are selling into a pool of buyers who are older, more cash-heavy, and fewer.

Two practical consequences. First, if your flat is approaching one of these thresholds, the cost of waiting another two years is not a smooth price taper — it is a step change in who can even bid. Second, if you are competing against a newer block nearby, the gap in months-of-inventory between you will be much larger than the gap in asking price suggests. I have written up the town-by-town lease profile separately at /hdb-lease-decay-by-town/, which is not live yet.

The five-day sale in Bedok

I have spent two thousand words telling you the market is slow. Now let me undercut myself, because a base rate is a description of a population, not a prophecy about a unit.

On 17 September 2026 — Bedok’s town record went to Blk 152B Bedok South Road at $1,540,000. That flat was listed on 12 September. It drew four enquiries within hours and it was reportedly sold in five days.

Five days, in a town where my own arithmetic says 4-room inventory runs at five months and the whole town turns over at 2.08% a year. Both facts are true at the same time.

The resolution is that the average is made of two populations. There is a population of flats that are correctly priced, well presented, in blocks with recent evidence, hitting the market when buyers who want exactly that thing are already looking. Those sell in days to weeks. And there is a population of flats priced against hope, in silent blocks, with photographs taken at dusk on a phone. Those sell in quarters, or not at all, and they are the ones dragging the mean.

The 6.7 months of national inventory is the weighted average of those two groups. You get to choose which group you are in. That choice is made on the day you set the price, not in month four when you cut it.

I should flag something about that Bedok sale, in the interest of being consistent about my own standards: it happened on 17 September 2026, and the public registry I have been quoting all the way through this article runs only to the end of August 2026. So that transaction is reported, not yet registry-verified. It will appear in the data release in a month or two. I am telling you it happened because I follow that town closely and I have written it up on my Bedok HDB resale 2026 page, but you should hold it a notch looser than everything above it. The same applies to the broader million-dollar flat trend, which I track in million dollar HDB flats 2026.

Who should expect a fast sale, who should not

Five ways sellers add months without noticing

1. Testing the market. Listing high “just to see” costs you the first fourteen days, which are the only days your listing is new. By the time you correct, you are competing as a stale flat against fresh ones.
2. Refusing viewings that are inconvenient. In a market where your block produces two or three sales a year, turning away a Tuesday evening viewing is turning away a meaningful fraction of your total annual buyer flow.
3. Waiting for the valuation before deciding on price. Valuation comes after the OTP. If you have not done your own comparable work before listing, you are negotiating blind.
4. Selling and buying in the wrong order without a plan. A seller with a hard deadline negotiates from weakness and every buyer can smell it. Work out the sequencing first — I cover it in the MOP guide, and there is a longer treatment coming at /sell-first-or-buy-first-hdb/.
5. Believing a seasonal story instead of the price. “We will wait for after Chinese New Year” costs you a quarter to chase a seventeen percent effect that will be swamped by getting the number right.

Where this analysis can be wrong

I would rather you find the holes here than in the comments.

The denominator is stale, and it cuts both ways. The dwelling-units table is a FY2021 vintage; my numerator is 2025 to 2026. Since 2021, HDB has completed and handed over a lot of new flats, so today’s true stock is larger than 1,032,699. A larger true denominator would make the real turnover rate lower than my 2.38%. Working the other way, many of those newly completed flats are still inside their five-year MOP and legally cannot be resold, so including them would understate the turnover of the flats that are actually sellable. My judgement is that the two effects partly cancel and 2.38% is close, but if you want a directional bet: the true all-stock rate is probably a touch below 2.38% and the true eligible-stock rate is probably a touch above. The town and flat-type comparisons are more robust than the absolute level, because the staleness applies to all cells.

Tengah is a hole. It shows zero stock in the FY2021 table because it had not been built yet. Any Tengah resales in my numerator would therefore have no denominator. There were essentially none in the window, so this does not bite, but it will within a couple of years.

Studio apartments and 2-room flexi muddy the small end. The stock table carries 8,708 studio apartments as a category the resale registry does not use, and 2-room stock includes flexi units under short leases with resale restrictions. My 2-room turnover figure of 2.34% should be read loosely.

Listing counts are portal data, not official data. One portal, one day, with duplicates and stale entries I cannot strip out. Every months-of-inventory figure inherits that.

Registration lag. Every month in the registry sits one to two months behind the handshake. That matters for the seasonality section and I have adjusted for it verbally, not statistically.

The summariser problem. I read these numbers through an automated layer that has been caught misreading digits before. Two of the twelve monthly counts disagreed between my first and second reads and needed a third before two agreed. Every published integer here survived at least two agreeing reads, and the two big arithmetic identities — the flat-type column summing to 24,590 and the seasonality months plus 2026 summing to 240,807 — both closed exactly. If they had not, I would have gone back rather than published.

FAQ