Singapore's Most Expensive Landed District Is the Only One Where Sellers Lose Money
There is a story Singapore's landed market tells about itself: buy a detached house, hold for two years, walk away with $3 million. The data from 3,488 landed transactions across Q3 2024 to Q2 2026 appears to confirm it. But pull one thread and the whole narrative unravels — because the most expensive district in that dataset, where the typical property costs $13.7 million, is the only one where sellers lose money at all.
The only district where landed sellers lose money is also the most expensive: D4 Sentosa Cove, average loss $616,000 on a $13.7 million property — while every other major district profits.
This article tests three questions in sequence. Is the "detached wins fast" headline real, or a data artefact? Is D4's loss pattern a fluke of small numbers, or something structural? And does the location evidence explain why Sentosa Cove specifically — not landed housing generally — consistently destroys seller wealth?
The headline that doesn't survive first contact with the data
Start with the apparent story. Across all 3,488 transactions, detached houses look dominant: the highest sticker price, the highest average profit, and — according to the summary figure — the shortest hold period.
| Type | Transactions | Typical price | Typical land size | Avg profit | Typical hold |
|---|---|---|---|---|---|
| Detached | 381 | ~$11.6M | ~6,381 sqft | $3,000,028 | 2 yrs |
| Semi-detached | 1,103 | ~$6.05M | ~3,422 sqft | $1,737,149 | 3.5 yrs |
| Terrace | 2,004 | ~$4.22M | ~1,893 sqft | $1,305,561 | 4 yrs |
Note: PSF for landed is calculated on land area, not strata floor area — not comparable to condo PSF figures.
The "detached wins fastest" reading seems clean. But now look at what happens when you sort the same dataset by hold period across all landed types:
| Hold period | Transactions | Avg profit | Win rate |
|---|---|---|---|
| Under 2 yrs | 45 | $1,630,298 | 76% |
| 2–4 yrs | 318 | $1,751,440 | 95% |
| 4–7 yrs | 333 | $1,966,075 | 99% |
| 7–10 yrs | 202 | $2,469,276 | 100% |
| 10+ yrs | 1,141 | $3,407,566 | 99% |
The relationship is completely linear. Hold longer, win more, win more often. The 10+ year cohort is the largest group in the dataset — 1,141 transactions — and averages $3.4 million in profit, which is actually higher than the national detached average of $3 million. The "short hold, big win" story for detached is sitting on top of a dataset where long holders are the dominant pattern and the biggest earners. These two tables should not be able to coexist the way they appear to. Something is wrong with one of them.
The 2-year hold figure for detached is almost certainly wrong
The transaction-level diagnostic resolves the contradiction. When you look only at detached sellers who have known profit and hold data — rather than rows where the hold period field is blank — a completely different picture emerges.
| Segment | Transactions | Known-profit rows | Known avg profit | Typical entry | Typical exit | Typical hold |
|---|---|---|---|---|---|---|
| D10 detached | 71 | 34 | $11,552,618 | ~$8.59M | ~$20.25M | 14.5 yrs |
| D10 semi-detached | 149 | 89 | $3,805,840 | ~$3.53M | ~$8.75M | 9.5 yrs |
| D10 terrace | 29 | 15 | $3,684,533 | ~$3.0M | ~$6.5M | 8.4 yrs |
| D11 all landed | 139 | 65 | $4,645,962 | ~$4.3M | ~$9.4M | 14.9 yrs |
D10 detached sellers with known profit data held for 14.5 years, not 2. D11 sellers held for 14.9 years. The national "2-year typical hold" for detached almost certainly reflects large volumes of blank holding-period fields in the summary data — not actual behaviour. Among sellers who have complete records, detached houses are held for well over a decade before sale.
This is not a caveat. It is the finding. The "short hold, big win" framing for detached is a missing-data artefact. The real story in Singapore's landed market is long-hold, district-specific, and heavily concentrated in a handful of streets in D10 and D11.
D10 detached sellers — 71 transactions, roughly 19% of all detached deals nationwide — averaged $5.53 million in profit per transaction. They are pulling the national detached average up considerably. Strip D10 out and the headline looks very different.
D4: Singapore's priciest landed district, its only losing one
Here is where the investigation turns. If long holds in D10 and D11 explain the top of the profit table, what explains the bottom? The answer is one district, and it inverts everything the market expects.
D4 has the highest typical transaction price of any district in the dataset — around $13.7 million per property. It also has the only negative average profit: -$616,094. Across 12 landed transactions, sellers on Sentosa island streets lost money consistently.
| Street | Transactions | Known-profit avg | Typical price | Notes |
|---|---|---|---|---|
| Ocean Drive | 4 | -$300,000 | ~$13.74M | Sentosa island |
| Cove Drive | 3 | -$1,365,000 | ~$14.16M | Sentosa island |
| Paradise Island | 1 | -$7,898,000 | $12.0M | Sentosa island |
| Cove Grove | 1 | n/a | $17.0M | Sentosa island |
| Treasure Island | 1 | +$4,899,872 | $17.8M | One profitable outlier |
| Wishart Road | 2 | n/a | ~$5.4M | Mainland — different sub-market |
Two clarifications matter here. First, Wishart Road is not Sentosa Cove. It is on the Singapore mainland with a 9-minute walk to Telok Blangah MRT, access to HarbourFront Centre, and bus connections. It is categorically a different product from the island streets. Its two transactions have no known profit data, but its location profile confirms it does not belong in the same analysis. D4's loss pattern belongs specifically to Sentosa island, not D4 broadly.
Second, the 12-transaction sample is small, and the district-level average of -$616,094 carries meaningful uncertainty. The one profitable outlier — Treasure Island at +$4.9 million on a single transaction — shows the average could shift materially with a handful of different sales. Treat the direction as clear; treat the exact figure as indicative.
The direction is clear because the private condo data says exactly the same thing.
Sentosa Cove condos confirm what the landed data shows
Private condo PSF is strata-based; landed PSF is land-based — they cannot be compared directly. But the direction of profit and loss can be compared, and here the two datasets align precisely.
| Project | District | Transactions | Avg loss | Win rate | Typical hold |
|---|---|---|---|---|---|
| Seascape | D4 | 11 | -$936,491 | 0% | ~3.7 yrs |
| The Coast at Sentosa Cove | D4 | 19 | -$464,712 | 16% | ~13.75 yrs |
| The Oceanfront @ Sentosa Cove | D4 | 22 | -$160,346 | 45% | ~13.7 yrs |
Seascape — 11 transactions, 0% win rate. Not a single profitable exit in the dataset. The Coast at Sentosa Cove: 19 transactions, only 3 in 19 sellers came out ahead. The Oceanfront, the most forgiving of the three, still shows less than half of sellers breaking even — after holding for nearly 14 years.
This is not a landed-specific phenomenon. Sentosa Cove is consistently loss-making across both asset classes, across multiple projects, across dozens of independent transactions. The combination of 12 landed transactions and 52 condo transactions all pointing the same direction moves this from a data curiosity to a structural pattern.
Sentosa Cove has no MRT, no hawker centres, no malls — and consistent losses
Why does this happen? The dataset cannot answer definitively — it cannot track buyer nationality, ownership intent, or the effect of foreign-ownership rules on resale demand. But the location data reveals something that is at least consistent with structural illiquidity.
| Street | Nearest MRT | Walk time | Hawker centres (500m) | Malls (1km) | Job nodes (2km) |
|---|---|---|---|---|---|
| Ocean Drive | HarbourFront (CC29) | ~69 min walk | None | None | None |
| Cove Drive | HarbourFront (CC29) | ~79 min walk | None | None | None |
| Paradise Island | Tanjong Pagar | ~128 min walk | None | None | None |
These are roughly approximate walking distances — treat them as directional — but the direction is unambiguous. Sentosa Cove is completely car-dependent. There are no hawker centres within 500 metres, no malls within 1 kilometre, no employment nodes within 2 kilometres. The nearest school found in the area is a single international preparatory school. There is no school catchment, no transit access, no daily infrastructure of any kind within reach.
When demand for an asset is entirely dependent on one thing — exclusivity as an end in itself — there is no fallback. If buyer sentiment toward that exclusivity cools, there is nothing else supporting the price. No transit catchment. No school premium. No employment-node logic.
Three competing explanations exist for why Sentosa Cove specifically underperforms, and the dataset cannot adjudicate between them:
Foreign-ownership rules. Sentosa Cove is one of the few places where foreigners could buy Singapore landed property. A large share of buyers may have purchased at the 2010–2013 price peak. Subsequent cooling measures and ABSD changes may have structurally constrained the resale buyer pool — foreigners who once drove demand face much higher costs to re-enter. The dataset cannot verify buyer nationality, but this mechanism is widely discussed and plausible.
Physical isolation. The location data is consistent with a buyer pool so narrow that when sentiment shifts, there is no residual demand driver to absorb supply. A 79-minute walk to the nearest MRT is not a minor inconvenience — it is complete car dependency on an island, which eliminates most Singapore buyers who rely on transit access.
Peak-cycle timing. With only 12 landed transactions across the full dataset period, D4's landed market is the thinnest in the dataset. It is possible that the loss pattern reflects a cohort of buyers who purchased at the 2011–2013 Sentosa Cove peak and are now selling into a permanently repriced market. A timing story and a structural story are not mutually exclusive — peak buying into a structurally illiquid market is worse than either alone.
The data cannot tell us which of these three is doing the most work. It can only tell us that losses are consistent, cross-confirmed, and geographically concentrated on Sentosa island specifically.
D10 sellers held 14.5 years, not 2 — and averaged $11.5 million in profit
The contrast with D10 is not subtle.
| D10 Duchess Avenue | Cove Drive | |
|---|---|---|
| Transactions | 6 | 3 |
| Typical price | ~$11.83M | ~$14.16M |
| Nearest MRT | ~12 min (Tan Kah Kee) | ~79 min (HarbourFront) |
| Schools within 1km | Nanyang Primary, Raffles Girls' Primary | None (local) |
| Malls within 1km | One Holland Village, ~0.9km | None |
| Known-profit avg | +$8,435,250 | -$1,365,000 |
Duchess Avenue costs less than Cove Drive. Its sellers average nearly $10 million more in profit per transaction. The premium that Cove Drive commands over Duchess Avenue at purchase is exactly inverted at sale. You pay more to lose more.
The D10 advantage is not unique to Duchess Avenue. It runs across the district's top streets:
| Street | Transactions | Typical price | Typical land | Known-profit avg |
|---|---|---|---|---|
| Duchess Avenue | 6 | ~$11.83M | ~3,965 sqft | $8,435,250 |
| Duchess Road | 5 | ~$11.38M | ~4,237 sqft | $7,440,000 |
| Namly Avenue | 7 | ~$8.88M | ~3,232 sqft | $6,162,164 |
| Grove Drive | 5 | ~$10.5M | ~3,940 sqft | $5,700,000 |
| Coronation Road West | 8 | ~$9.71M | ~4,210 sqft | $4,081,000 |
Every one of these streets sits within reasonable distance of an MRT station, within catchment of established primary schools, and near retail and daily-life infrastructure. The profit is not from detached housing as a type — it is from this combination of location depth and long holding periods.
D11 (Newton / Novena / Watten) tells a similar story: 65 known-profit rows, 14.9-year typical hold, $4.65 million average profit. Kheam Hock Road leads D11 at $8.595 million average profit across 8 transactions, with roughly an 18-minute walk to Botanic Gardens MRT.
A D10 terrace outearns the average detached house nationwide
One finding sits outside the main investigation but earns its own mention. D10 terrace sellers with known profit data averaged $3,684,533 — higher than the national average for all detached houses at $3,000,028.
A terrace house in D10 outperforms the average detached house anywhere else in Singapore. The type label does not determine the outcome — the postcode does.
What three years of landed data actually shows
Strip away the composition effects and the missing-data artefacts, and Singapore's landed market reduces to one clear pattern: district and hold period explain almost everything.
"Detached wins, holds short" is not a real finding. It is what you see when a dataset has large numbers of blank holding-period fields and one district — D10 — pulling the average profit figure up for an entire property type.
The one place in Singapore where the "expensive landed, short hold" combination actually occurs is Sentosa Cove. And it is the one place where sellers consistently lose money. The properties that averaged the highest losses — Cove Drive at -$1.37 million, Paradise Island at -$7.9 million per known transaction — are also among the highest-priced properties in the entire dataset.
The private condo market on the same streets, across 52 independent transactions, produces the same answer.
What the data cannot say is why Sentosa Cove loses while every other major district wins. The three plausible mechanisms — foreign-ownership rules, physical isolation, and peak-cycle timing — are not mutually exclusive, and the dataset has no field that distinguishes between them. The location data is consistent with structural illiquidity. The timeline is consistent with a peak-buying cohort selling at a loss. The ownership rules are consistent with a permanently narrowed resale buyer pool. All three could be true simultaneously.
What the data can say is that paying the highest price in Singapore's landed market has not, in this dataset, produced the highest returns. It has produced the only losses.
For more on how composition effects distort district-level averages in Singapore property, see D10's $3.16M Average Is a Detached-House Story. Semi-Detached Buyers Are Reading the Wrong Number.