95.5% of 3 bedroom owners who sold since 2020 sold for more than they paid, with a median gain of $455,000

Between January 2020 and August 2026, 28,614 3 bedroom condominiums changed hands where the same unit had been bought at a recorded price earlier. Of those, 27,318 sold for more than the owner paid, and 1,261 sold for less.

That is 95.5% above water, before any transaction cost. The median gain was $455,000, or 34.0%, over a median hold of 9.6 years, and the median yearly return was 3.5%.

The medians hide two things worth knowing: where the losses are, and when the bad years to buy were.

What is in this note

• What owners made: how many 3 bedroom resales sold above what the owner paid, and by how much.
• The CCR has the largest gains and the highest chance of a loss: the same figures split by region.
• When you bought mattered more than where: annualised return by the window the owner bought in, and how that compares with the spread between regions.
• The spread: what the best and worst tenth of owners made, and how far apart they are.

What owners made

Pairs measured	28,614
Sold above	27,318 (95.5%)
Sold below	1,261 (4.4%)
Median gain	$455,000
Median gain, per cent	34.0%
Median hold	9.6 years
Median return per year	3.5%

A further 35 units sold for exactly what the owner paid, which is why the two shares above do not sum to 100%.

A 3 bedroom held for the better part of a decade made its owner about a third more than they paid. That is the central case, and it is not a dramatic number: 3.5% a year, before any of the costs of owning.

The CCR has the largest gains and the highest chance of a loss

Region	Sold above	Median gain
Core central (CCR)	90.1% (3,876)	$620,950 (27.7%)
Rest of central (RCR)	95.1% (9,206)	$510,000 (34.0%)
Outside central (OCR)	97.0% (15,532)	$405,000 (35.5%)

Region	Hold	Per year
Core central (CCR)	11.1 years	2.71%
Rest of central (RCR)	9.1 years	3.78%
Outside central (OCR)	9.5 years	3.59%

In dollars the CCR is ahead: $620,950 against $405,000 in the OCR. In the chance of a gain it is behind: 90.1% of CCR owners sold above what they paid, against 97.0% in the OCR. So roughly one CCR owner in ten sold at a loss, and when they did the median loss was $320,000.

Both facts follow from price. A CCR 3 bedroom costs more, so both the gains and the losses are larger in dollars, and the percentage return is the lowest of the three regions at 2.71% a year.

When you bought mattered more than where

Bought	Sold above	Median gain
Before 2010	97.0% (7,721)	$846,640 (98.8%)
2010 to 2014	92.4% (9,756)	$345,000 (26.6%)
2015 to 2018	96.6% (6,427)	$370,000 (26.5%)
2019 to 2022	98.3% (4,559)	$406,888 (26.7%)

Bought	Hold	Per year
Before 2010	17.7 years	3.87%
2010 to 2014	10.4 years	2.34%
2015 to 2018	6.1 years	4.05%
2019 to 2022	4.0 years	5.98%

Of the four purchase windows, 2010 to 2014 is the outlier. Owners who bought in that window were least likely to be above water at 92.4%, and made the least per year at 2.34%. Every other cohort returned between 3.87% and 5.98%.

That period was the top of the last cycle. The total debt servicing ratio framework arrived in June 2013, and the additional buyer's stamp duty had been introduced in December 2011 and raised in January 2013. Buyers who entered just before those measures paid prices the market took years to reach again.

The cohort bought since 2023 is not shown. It holds 151 pairs, but the units in it have been held for a median of 3.0 years, too short a hold to set beside the windows above.

The spread

A single median hides a wide spread between owners.

	Gain
Worst tenth	under +7.3%
Lower quarter	under +18.1%
Median	+34.0%
Upper quarter	over +62.0%
Best tenth	over +120.7%

The gap between the tenth and ninetieth percentile is 113 percentage points. Two owners buying the same year in the same region can land anywhere in that range depending on the project, the unit and when they sold.

And for the 4.4% who lost money, the median loss was $130,000, or 6.7%. The worst tenth of those lost more than 23.0%.

What this tells you

Of 28,614 3 bedroom resales since January 2020 where the same unit had a recorded earlier purchase, 27,318 sold above what the owner paid and 1,261 sold below. That is 95.5% above water, before any transaction cost, with a median gain of $455,000 or 34.0% over a median hold of 9.6 years, and a median yearly return of 3.5%.

The 4.4% who sold below paid is not evenly spread. In the CCR 9.8% of owners sold below the purchase price against 2.8% in the OCR, so roughly one owner in ten in the CCR took a loss against one owner in thirty-five in the OCR.

The CCR is also where the largest gains are. A CCR owner's median gain was $620,950 against $510,000 in the RCR and $405,000 in the OCR. As a return per year it inverts: the CCR returns 2.71% a year, the lowest of the three regions, against 3.78% in the RCR and 3.59% in the OCR. The CCR owner made larger sums on larger, longer-held assets, at a lower rate.

When the owner bought mattered more than where. Across the three regions annualised return spans 1.07 percentage points, from 2.71% to 3.78%. Across purchase windows it spans 3.64 points, and that span is not evenly filled: three of the four windows sit between 3.87% and 5.98%, and one window, 2010 to 2014, sits alone at 2.34%. Timing moved returns 3.4 times as much as region did, and most of that is one bad window rather than a steady change across the purchase windows.

The widest spread of all is between individual owners, and it is not explained by region or timing. The worst tenth made under 7.3% in total and the best tenth over 120.7%, a gap of 113 points. Two owners who both bought a 3 bedroom and both sold at a profit could have made 7% or 121%, and the median of 34.0% sits between two very different outcomes.

What this does not tell you

• Whether your timing matches theirs. The 2010 to 2014 cohort shows the result of buying at the wrong moment: 2.34% a year, the lowest of the four purchase windows.
• Your costs. The gain is the difference between two prices. Stamp duty, agent fees, legal costs, property tax, maintenance and any interest paid are not in it, and over a decade they are substantial.
• What owners who did not sell have. Only units that transacted twice appear here. Anyone still holding is invisible, which means this is a measure of realised outcomes rather than of the whole market.
• What any one development did. Every figure here pools 28,614 resales across every 3 bedroom development, and this note measures none of them on its own. Owners spread widely around the median: the worst tenth gained under 7.3% and the best tenth over 120.7%.

Method and data notes. Sale data to 31 August 2026.

Matched pairs from URA caveat data: a 3 bedroom private condominium or apartment bought at a recorded price and resold, with the sale completing between 1 January 2020 and 31 August 2026. Executive condominiums are excluded. Bedroom count is taken from the purchase leg. Pairs are formed from consecutive transactions of the same unit.

Gains are the difference between the two recorded prices, before any transaction cost. The return per year is computed on pairs held at least 6 months. Units with an implausible area for their bedroom count are excluded: 111 pairs on the national figure. A further exclusion covers units found to be mislabelled on floor-plan review: two corpus-wide, of which one unit forms a pair in this national figure. Seven developments whose bedroom labels in the source records are under review are left out of every figure here, and so are the Reflections at Keppel Bay units whose recorded size does not fit the bedroom count.

Figures were computed on data as at 31 August 2026. Caveats lodged since then are not included, and the same figures back the 2 bedroom equivalent of this note, where 92.9% of owners sold above purchase for a median gain of $231,000 over 8.2 years, so the two can be compared directly. The tenth- and ninetieth-percentile gain figures are computed to four decimal places; the gap between them is computed from those values and rounded once for display, so it will not match arithmetic done on the one-decimal figures shown in the table above.

No figure resting on fewer than 100 pairs is published. The 2023-onward purchase cohort holds 151 pairs and is named but not tabulated; crossing region and purchase period produces groups as small as 13, and none of those appears here.

Sale data to 31 August 2026.

30 October 2026. This note placed homes in market segments by postal district. URA places them by planning area, which moves about one in seven private sales between regions. The figures by region here use the earlier segments and differ from what the same records give now. How regions are set: How we measure prices and sales.