A Sentosa Cove house loses money less often than an apartment. The typical house seller still lost $390,444.

A Sentosa Cove house loses money less often than a Sentosa Cove apartment. 53.7% of resold houses were sold below their purchase price, against 66.4% of apartments.

The typical house seller still lost $390,444. The typical apartment seller lost $300,000.

Both of those are true. Which one matters depends on whether you are counting percentages or dollars.

Key takeaways

• Houses lose less often, 53.7% against 66.4%, and by a smaller share of the purchase price. Which loses more often
• When a house wins, it wins large. Median gain $4,450,000 against $373,360 for an apartment. The winners
• When a house loses, it loses $2,659,558 at the median against $715,000 for an apartment. The losers
• Annualised, houses win faster and lose faster. +3.19% a year against +1.17% among the winners; -2.48% against -1.56% among the losers. Annualised, houses move faster
• Almost all of the headline advantage comes from houses being bought later. Hold the purchase year constant and it mostly disappears. Most of the gap is timing

Which loses more often

Resale pairs since March 2020. A pair is a home matched to its own previous sale, so both figures describe the same property changing hands twice.

	Houses	Apartments
Resale pairs	54	268
Sold below purchase price	53.7%	66.4%
Median outcome	-3.9%	-9.0%
Median outcome in dollars	-$390,444	-$300,000
Median purchase price	$14,150,000	$3,522,100

The percentage columns favour houses. The dollar column does not, and the reason is in the last row: the median house cost four times the median apartment, so a smaller percentage shortfall is more money.

The winners

	Houses	Apartments
Sold above purchase price	44.4% (24)	33.2% (89)
Median gain	$4,450,000	$373,360
Middle half of gains	$2.57m to $5.78m	$163,000 to $694,000
Largest gain	$11,666,000	$2,800,000
Median gain as a percentage	+30.9%	+13.5%

This is where the two segments differ most. A house that made money made a median $4.45 million, which is about twelve times an apartment's $373,360, and it did so on a purchase price only four times larger. In percentage terms the gap is 30.9% against 13.5%.

The largest single house gain in the period was $11,666,000, on a sale held under two years. The largest apartment gain was $2,800,000. The best percentage return on a house was 102.9%, on a different house that gained $3,550,000 over fifteen years; the best on an apartment was 99.0%.

The losers

	Houses	Apartments
Sold below purchase price	53.7% (29)	66.4% (178)
Median loss	-$2,659,558	-$715,000
Middle half of losses	-$4.70m to -$2.00m	-$1.63m to -$302,000
Largest loss	-$10,650,000	-$6,169,600
Median loss as a percentage	-19.6%	-18.3%

In total percentage the two look almost identical: a losing house sold 19.6% below its purchase price, a losing apartment 18.3%. That similarity is misleading, because houses were held for less time.

Annualised, they separate. A losing house fell 2.48% a year, a losing apartment 1.56%. So houses do not fall more gently. They fall faster, and from four times the height. A quarter of losing house sales were down more than $4.7 million, and the worst was down $10,650,000.

Annualised, houses move faster

Total returns are not comparable when holding periods differ, and here they do: houses were held a median 10.1 years, apartments 13.7. Compounded annual rates strip that out.

Median annual return	Houses	Apartments
All resales	-0.37%	-0.82%
Those that made money	+3.19%	+1.17%
Those that lost money	-2.48%	-1.56%

Houses are ahead overall and ahead among the winners by nearly two points a year. They are also behind among the losers by nearly a point.

That is the whole comparison in one line. A Sentosa Cove house wins more often, wins much faster, and when it goes wrong it goes wrong faster too.

Most of the gap is timing

The obvious next question is whether houses are simply the better asset here. Mostly not.

House sellers in this window bought later than apartment sellers, around 2013 against 2009 to 2010, and our earlier work found that when a Sentosa buyer bought is one of the two things that most determined the outcome. Later buyers did better in both segments.

Holding the purchase year constant, houses lead in seven of the nine years where both had sales. But the margin comes from 2006 and 2007, which rest on one and four house sales. In the three years with usable samples the gap is smaller: 3.3 percentage points of total return in 2010, 7.7 in 2011 and 2.0 in 2012. In 2013 it reverses, with apartments ahead by 3.8.

So the honest reading is that houses did modestly better within the same purchase year, not dramatically. The pooled gap of 12.7 percentage points on loss rate overstates it, because it is comparing sellers who bought in different markets.

What this means

For an owner comparing to the headline, the segment number is the wrong number. Whether you are ahead or behind depends far more on the year you bought than on whether you own a house or an apartment.

For anyone choosing between the two, the difference is in how far the outcomes range on each side. A house here has produced bigger wins and bigger losses, and annualised it moves faster in both directions.

A smaller percentage can still be the larger loss. A 3.9% shortfall on a $14 million house is $390,444. A 9.0% shortfall on a $3.5 million apartment is $300,000. The apartment looks worse as a percentage and costs less in money.

What this does not tell you

• What separates the houses that won from the ones that lost. We tested purchase year, holding period and plot size. None separates them, which is not the case for apartments, where the project and the purchase year both do. That is the subject of a separate note in this series.
• Street-level differences. Only two of the nine house streets have enough resales to read: Ocean Drive at 15 pairs and Cove Drive at 13, and they are close. The rest run from one to seven pairs. Sandy Island is worth flagging rather than concluding: five resales, all five below purchase price.
• Costs. Every figure is gross of stamp duties, agent fees, legal costs and holding costs. For a foreign buyer since April 2023 the stamp duty alone would change these outcomes materially.
• Anything about the wider island. This is the Sentosa Cove enclave only.

Method

Figures are transacted prices from URA caveat data for Sentosa Cove, excluding collective sales.

Houses are landed transactions on the nine Sentosa Cove residential streets: Ocean Drive, Cove Drive, Paradise Island, Coral Island, Treasure Island, Sandy Island, Cove Grove, Cove Way and Pearl Island. Jalan Sentosa and Sandy Lane are on the mainland despite their names and are excluded. Of the 54 house pairs, 50 are detached houses and 4 are terrace houses.

Apartments are the non-landed projects used in our earlier Sentosa Cove work.

A resale pair matches a home to its own previous recorded sale. For apartments the match is on the unit record. For houses it is on street and house number, because landed transactions carry no unit identifier. The outcome is the second price divided by the first.

Only pairs whose sale falls on or after 1 March 2020 are counted, so these figures describe recent sellers.

Medians are used rather than averages. The middle-half ranges are the 25th to 75th percentiles.

Largest gain and largest loss are the biggest single amounts in dollars. The best and worst percentage returns come from different sales, and are stated separately where they appear.

Sale data to 4 August 2026.