Every resale at Turquoise lost money. At The Berth by the Cove, the median seller made 5.1%. Both are Sentosa Cove.

Every resale at Turquoise since 2020 lost money. At The Berth by the Cove, the median seller made 5.1%.

Both are Sentosa Cove. Both sold into the same market. Their buyers bought in the same years, held for about the same time, and read the same headlines about the enclave. The difference between them is forty-six percentage points.

Key takeaways

• Median outcomes by project run from +5.1% to -41.7%. Turquoise lost money on every resale; The Berth lost on 42.9%. What each project did
• The ordering holds in every one of seven purchase years. It is not a timing effect wearing a project label. Holding the purchase year constant
• When you bought matters just as much and separately. A 2006 buyer at The Oceanfront is up 23.5%; a 2007 buyer at the same project is down 23.7%. One year apart
• Unit size explains almost nothing once the project is held constant. What size does not explain

What each project did

Resale pairs since March 2020, non-landed Sentosa Cove, projects with at least ten:

Project	Pairs	Median outcome	Loss-making	Median hold
The Berth by the Cove	49	+5.1%	42.9%	12.3 yrs
The Oceanfront	82	-3.9%	61.0%	13.7
The Azure	22	-9.5%	59.1%	12.8
The Coast	62	-10.0%	72.6%	14.4
Seascape	17	-30.1%	94.1%	11.5
Marina Collection	22	-32.4%	90.9%	13.7
Turquoise	13	-41.7%	100%	15.0

The top and bottom of that table are not describing the same investment. One project returned a small profit to the typical seller. Another returned a loss to every single one.

Holding periods are close: 11.5 to 15.0 years across the whole table. Nobody here was trading. These are long holds that ended differently.

Holding the purchase year constant

The obvious objection is that the weak projects were simply bought at worse moments. Our earlier work found purchase year drives Sentosa outcomes heavily, so this needs testing rather than assuming.

It is not the explanation. Median outcome by the year the seller bought:

Bought	The Berth	The Oceanfront	The Coast	Seascape, Marina, Turquoise
2006	+20.1%	+23.5%	-0.9%	no sales
2007	-17.4%	-23.7%	-21.0%	-44.1%
2008	-6.5%	-1.4%	-25.1%	-34.3%
2009	+6.4%	-3.9%	-10.9%	-18.0%
2010	-14.7%	-16.7%	-28.3%	-34.1%
2011	+4.3%	no sales	-21.9%	-38.5%
2012	-21.1%	-13.2%	-14.9%	-32.8%
2013	+3.7%	-6.6%	-10.2%	-25.7%

The three weakest projects are last in all seven years where they traded, by fifteen to twenty-five points. Seven out of seven is not a run of bad luck arranging itself.

Individual cells here rest on very few sales, some on one or two, and none of them should be read on its own. The claim that survives is the ordering, which does not depend on any single cell.

One year apart

The same table says something else worth pausing on.

At The Oceanfront, a seller who bought in 2006 is up 23.5%. A seller who bought in 2007 is down 23.7%. Same project, same building, one year between the two purchases, and forty-seven points between the outcomes.

Both effects are real and they are separate. Which project you bought set a floor and a ceiling. Which year you bought moved you within them, and the movement is as large as the spread between projects.

What size does not explain

Pooled across the enclave, two-bedrooms look like the only size where the median seller made money, and the loss rate appears to climb steadily with each additional bedroom. That reading does not survive.

Within a single project, by size:

Project	2-bed	3-bed	4-bed and up
The Berth by the Cove	+5.0%	+10.5%	-7.8%
The Oceanfront	-2.0%	-4.1%	-1.2%
The Azure	-6.2%	-14.7%	-11.6%
The Coast	no sales	-9.6%	-12.5%
Seascape	no sales	-30.1%	-26.2%
Marina Collection	no sales	-32.8%	-31.8%
Turquoise	no sales	-41.7%	-35.1%

Sizes do not sum to the project totals above. A small number of sales carry no resolved bedroom count in the unit record and are excluded from this table.

There is no consistent pattern. In three of the seven, the larger unit did better. At The Berth the three-bedroom beat the two-bedroom.

The pooled size ladder was a project effect in disguise. Two-bedrooms exist in only three of these projects, and those three are near the top of the table. The apparent size finding was measuring which project happened to be selling.

What this means

A single number for Sentosa Cove hides more than it says. The enclave-wide loss rate reported by The Business Times in August 2026, and the figure we reproduced ourselves in our earlier note, is an average across projects whose outcomes differ by forty-six points. It describes no actual property.

For an owner, the relevant comparison is inside your own project. What the enclave did tells you little about what your building did. Our project-level medians differ from the enclave median by up to thirty points in either direction.

For anyone valuing a Sentosa unit, project and purchase year are the two things that carry the information. Size did not survive the test. Whatever separates these projects, it is stable enough to hold across seven purchase years, which is a stronger regularity than most things in this market.

What this does not tell you

• Why the projects differ. We can measure that they do and that it persists across purchase years. Caveat records do not say whether it is the marina berths, the sea frontage, the tenure structure, the maintenance, the developer, or which units happen to reach the market.
• Anything about the smaller projects. Nine of the nineteen Sentosa Cove projects have fewer than ten resale pairs since 2020 and are excluded. They may behave differently.
• The landed enclaves. This is non-landed only. Sentosa Cove has roughly 392 landed homes and they are not in these figures.
• What any single unit is worth. These are project medians over long holds. Floor, facing, berth access and condition matter for a specific home in ways a project median cannot capture.

Method

Figures are transacted prices from URA caveat data for non-landed homes in Sentosa Cove, excluding collective sales.

A resale pair matches a unit to its own previous recorded sale. The outcome is the second price divided by the first, gross of stamp duties, agent fees, legal costs and holding costs. This is the same method as our earlier work on the enclave.

Only pairs whose sale falls on or after 1 March 2020 are counted, so the table describes recent sellers rather than the whole history.

Projects with fewer than ten such pairs are excluded from the project table. In the purchase-year and size tables, Seascape, Marina Collection and Turquoise are grouped where individual cells would otherwise rest on one or two sales.

Bedroom count is worked out from the unit record rather than stated in the caveat.

Medians are used rather than averages.

Sale data to 4 August 2026.