The Three Projects That Lost the Most Money Were All Large Developments. The Riskiest Condo Is Not What You Have Been Told.

Atlas compares private condo markets where resale profits reverse.

A 265-unit leasehold condo in Geylang posted a 100% win rate across 31 resale transactions. Meanwhile, a 1,042-unit development in the heart of the city returned a 3.6% win rate — meaning 96 out of every 100 sellers lost money. The small project is in District 14. The large one is in District 1. If boutique condos are supposed to be the riskier bet, something has gone very wrong with that theory.

This article tests the claim directly: are boutique condos actually riskier to hold than large ones? The data does not give a clean verdict. But it does show clearly which variable actually drove the worst outcomes — and it is not project size.

The three worst-performing projects in this dataset — Marina One Residences (3.6% win rate, −1.9% p.a.), OUE Twin Peaks (7% win rate, −1.71% p.a.), and Reflections At Keppel Bay (50.9% win rate, +0.35% p.a. after a decade) — are all large-to-mega scale CCR leasehold developments. The separating variable is when buyers entered and at what price — not how many units the project has.

One caveat before the numbers: unit count figures throughout this article are approximate. The resale data does not confirm exact developer counts. Treat all project sizes as directional.


The Liquidity Gap Is Real — And Worth Stating Plainly

Before testing whether boutiques are riskier, the honest answer to the first question: yes, they genuinely move fewer units per year.

Project Scale (approx.) Transactions Avg hold (yrs) Approx. resales/yr
Stirling Residences ~1,259 units (mega) 211 5.70 ~37
Reflections At Keppel Bay ~1,129 units (mega) 118 10.66 ~11
Marina One Residences ~1,042 units (mega) 56 7.02 ~8
The Antares ~265 units 31 4.31 ~7
Sky Everton ~262 units 19 5.59 ~3
Skyline Residences ~283 units 29 9.98 ~3
Espada ~59 units 16 12.69 ~1
Helios Residences small CCR 10 12.58 ~1

Stirling Residences (Queenstown, roughly 6 minutes from Queenstown MRT) generates around 37 resale transactions a year. Espada on St Thomas Walk generates around one. That gap is real, and anyone selling a boutique unit needs to plan around it: fewer buyers will see the listing, and timing matters more.

But the per-year figures above also contain a warning about the other side of the table. Marina One Residences — a 1,042-unit development — generates only around 8 resale transactions per year. That is barely more than The Antares, a 265-unit project. The reason is not that the project is illiquid. It is that most sellers who bought at peak prices are not willing to crystallise a loss unless they have to. A loss-making mega project and a boutique project can look equally thin in the resale data — for completely different reasons.


Three Large Projects, Three Poor Outcomes

The worst composite performers in this dataset are not what the conventional risk narrative predicts.

Project District Tenure Scale (approx.) Transactions Win Rate Avg return p.a.
Marina One Residences CCR / D1 99-year ~1,042 units 56 3.6% −1.9%
OUE Twin Peaks CCR / D9 99-year ~462 units 43 7.0% −1.71%
Reflections At Keppel Bay RCR / D4 99-year ~1,129 units 118 50.9% +0.35%

Marina One Residences (D1, completed 2017) sold to buyers largely during the 2012–2014 launch window at peak CCR pricing. Across 56 resale transactions, 96.4% lost money. The annualised loss averaged −1.9% per year.

OUE Twin Peaks (D9 River Valley, completed 2015, ~462 units) is not a mega project — it sits in the mid-scale range. But it produced a 7% win rate and −1.71% p.a. across 43 transactions. That performance is worse than most boutique freehold projects in this dataset, and it dispels any idea that safety lives somewhere between the two size extremes.

Reflections At Keppel Bay (Keppel Bay View, completed 2011, ~1,129 units) tells a more nuanced story. The average loss across all 118 transactions was roughly $84,000. But the median profit — the figure closest to a typical seller's experience — was near $8,150. A small number of buyers who entered at the 2011 launch peak dragged the average into loss territory; most sellers barely broke even. After a decade of holding, that is a poor outcome by any measure, but it is not the uniform disaster the average suggests.

The shared characteristic across all three: CCR leasehold, 99-year tenure, and purchase windows clustered around the 2011–2015 CCR pricing peak.


Same District, Three Projects, One Clean Comparison

The clearest test of whether project size drives resale risk is in District 4, where three projects with comparable unit sizes and completions within five years of each other tell three different stories.

Skyline Residences Corals At Keppel Bay Reflections At Keppel Bay
Location Telok Blangah Road Keppel Bay Drive Keppel Bay View
Nearest MRT Telok Blangah, ~8 min Harbourfront, ~10 min Telok Blangah, ~11 min
Tenure Freehold 99-year 99-year
Completed 2015 2016 2011
Scale (approx.) ~283 units ~200 units ~1,129 units
Typical unit (sqft) ~1,292 ~1,238 ~1,346
Transactions 29 32 118
Win Rate 75.9% 43.8% 50.9%
Avg return p.a. +0.88% −0.12% +0.35%
Avg hold (yrs) 9.98 8.32 10.66

Same district. Similar unit sizes — roughly 1,238 to 1,346 sqft. All three completed between 2011 and 2016. If project size were the key variable, the largest project should outperform the smallest. Instead, the freehold mid-scale project (Skyline Residences, ~283 units) outperformed both the smaller leasehold project and the larger leasehold mega. Corals At Keppel Bay averaged a nominal loss across 32 transactions. Reflections barely broke even after more than a decade.

The isolating variable is tenure — not unit count. Both leasehold projects underperformed regardless of whether they were boutique-scale or mega-scale.


Vintage Matters as Much as Tenure — The D9 Comparison

The D4 matched pair identifies tenure as a separating variable. But District 9 tests something sharper: does CCR leasehold itself predict failure, or is it specifically about when you bought?

OUE Twin Peaks Martin Modern
District CCR / D9 CCR / D9
Tenure 99-year 99-year
Scale (approx.) ~462 units ~450 units
Completed 2015 2021
Transactions 43 56
Typical unit (sqft) ~1,055 ~867
Typical PSF $2,241 $2,748
Win Rate 7.0% 82.1%
Avg return p.a. −1.71% +1.58%

Both are CCR D9 leasehold projects. Similar unit counts. Martin Modern (D9 / River Valley, ~450 units, completed 2021) produced an 82.1% win rate and +1.58% p.a. across 56 transactions. OUE Twin Peaks — same region, same tenure structure, similar scale — produced a 7% win rate and −1.71% p.a.

The difference is not tenure. It is not region. It is not project size. Buyers who entered OUE Twin Peaks during the 2012–2015 CCR pricing peak paid prices that the subsequent decade could not recover. Buyers who entered Martin Modern on a pre-2021 new-launch basis bought before the post-pandemic run-up and sold into a stronger market. When you bought, not how many units the project has, determines whether capital can be recovered.

D'Leedon (CCR D10, ~1,703 units, completed 2014, 99-year) adds another data point in the same direction: 95.9% win rate and +2.9% p.a. across 122 transactions — a large CCR leasehold project that performed strongly. The typical resale PSF at D'Leedon ($2,053) and the long average hold periods suggest buyers who entered earlier or at lower initial prices. The CCR leasehold mega failure narrative is real, but it is specific: it applies to the 2011–2015 peak-entry cohort, not to the category as a whole.


The Projects That Actually Did Well Were Small and Newer

While the large CCR leasehold projects from the 2011–2015 cohort were struggling, several smaller and newer projects were producing results at the other end of the table.

Project Region / District Tenure Scale (approx.) Transactions Typical unit (sqft) Win Rate Avg return p.a.
The Antares RCR / D14 99-year ~265 units 31 ~710 100% +3.3%
Forett@Bukit Timah RCR / D21 Freehold ~633 units 26 ~732 100% +3.2%
Sky Everton RCR / D2 Freehold ~262 units 19 ~624 100% +1.7%
Riviere RCR / D3 99-year ~455 units 18 ~991 94.4% +2.1%
Cyan CCR / D10 Freehold small 16 ~1,001 87.5% +2.0%
Skyline Residences RCR / D4 Freehold ~283 units 29 ~1,292 75.9% +0.88%

The most counterintuitive entry is The Antares. A 99-year leasehold project in District 14 — Geylang planning area, Mattar Road — produced a 100% win rate across 31 transactions and averaged +3.3% per year. The conventional risk checklist would have flagged this: leasehold, not CCR, not a prestigious address. And yet it outperformed developments in D1, D4, and D9 with freehold title and larger unit counts.

Location and entry timing explain most of it. The Antares sits about 5 minutes from Mattar MRT on the Downtown Line — the best MRT access of any project in this comparison. It completed in 2022, meaning buyers entered at pre-completion new-launch pricing rather than at a pricing peak. The occupier catchment along the Mattar-MacPherson-Paya Lebar corridor is genuine.

Forett@Bukit Timah (D21, freehold, 633 units, completed 2024) is worth naming accurately: at roughly 633 units, it is moderate scale rather than boutique. Its 100% win rate across 26 transactions reflects both freehold tenure and a very short holding window post-completion — treat that as directional, not a long-term verdict. Sky Everton (Everton Road, ~262 units, freehold, completed 2022) and Riviere (1 Jiak Kim Street, Robertson Quay waterfront, ~455 units, 99-year, 6 minutes from Havelock MRT on the Thomson-East Coast Line) both produced 100% and 94.4% win rates respectively on thin transaction counts — the direction is clear but 18–19 transactions means these are signals, not conclusions.

The shared conditions across the outperformers: RCR location, completions from 2015 onwards, and real occupier sub-markets with MRT access.


Not All Small CCR Projects Are the Same Failure

The boutique underperformer section splits into two distinct patterns, and conflating them produces a misleading picture.

Pattern one — CCR micro-unit freehold: Espada (St Thomas Walk, ~59 units, ~560 sqft typical unit, 12.5% win rate, −0.28% p.a. across 16 transactions) and 26 Newton (D11 Novena, ~560 sqft typical unit, 21.1% win rate, −0.74% p.a. across 19 transactions). These are genuine micro-unit CCR freehold projects where entry PSF was high, buyer pool was narrow, and resale demand never materialised to the level needed to recover capital. The risk here is unit size and buyer-pool depth, not project size.

Pattern two — large-unit peak-era CCR freehold: Helios Residences (15 Cairnhill Circle, D9, freehold, ~1,491 sqft typical unit, 0% win rate, −1.9% p.a. across 10 transactions) is a different failure entirely. These are large apartments — bigger than most units in this comparison — bought at or near the 2011 CCR completion cycle on Cairnhill Circle, roughly 16 minutes from Newton MRT. The loss is driven by peak-era entry pricing on large units in a CCR pocket that never recovered those levels, not by illiquidity from small unit size. With only 10 transactions, the Helios result is directional rather than definitive — but the direction is unambiguous.

Espada rents at $8.57 per sqft, which is the highest in this comparison and suggests genuine demand for the River Valley micro-unit format from tenants. That rental premium does not offset the resale loss, but it does mean the holding experience was not entirely painful. Helios Residences, by contrast, rents at $5.44 per sqft — below Marina One ($6.64/sqft) and barely above Reflections ($5.39/sqft). For Helios, the rental market offers no compensation for the resale weakness.

The lesson: CCR boutique does not equal one risk type. The failure mode for a 560-sqft River Valley unit is structurally different from the failure mode for a 1,491-sqft Cairnhill Circle apartment.


Where the Rental Scarcity Premium Shows Up — and Where It Does Not

Project Scale Rental rows (2 yr) Typical rent PSF
Espada ~59 units 267 $8.57
Riviere ~455 units 235 $8.26
Sky Everton ~262 units 168 $7.35
Stirling Residences ~1,259 units 761 $7.05
Spottiswoode Suites ~43 units 178 $7.27
26 Newton boutique 160 $6.67
Marina One Residences ~1,042 units 998 $6.64
Corals At Keppel Bay ~200 units 236 $6.35
Normanton Park ~3,000 units 1,308 $6.11
Parc Clematis ~1,468 units 520 $6.15
Helios Residences small CCR 71 $5.44
D'Leedon ~1,703 units 898 $5.45
Reflections At Keppel Bay ~1,129 units 630 $5.39

The rental scarcity argument holds for the right boutiques. Espada at $8.57/sqft, Riviere at $8.26/sqft, and Sky Everton at $7.35/sqft all sit comfortably above the mega-project range. Stirling Residences — one of the strongest resale performers in the mega category — rents at $7.05/sqft, which is notable for a 1,259-unit project and reflects the depth of the Queenstown rental market rather than any scarcity effect.

But Helios at $5.44/sqft and D'Leedon at $5.45/sqft show that neither boutique scale nor CCR address guarantees a rental premium. The premium tracks occupier-market depth — River Valley, Robertson Quay, and the Tanjong Pagar corridor generate it. Cairnhill Circle and Keppel Bay do not, at least not at these price points.


What the Data Actually Shows — and What It Leaves Open

The liquidity claim against boutique condos is real. A project with 59 units will generate roughly one resale transaction per year. That is a genuine constraint on exit timing and buyer pool size.

But the price-performance claim runs in the opposite direction to the conventional warning. The worst outcomes in this dataset came from CCR leasehold projects — large, mid-scale, and boutique alike — where buyers entered during the 2011–2015 pricing peak. Scale offered no protection. A 1,042-unit project (Marina One) and a 462-unit project (OUE Twin Peaks) both produced single-digit win rates in the same CCR leasehold category. A 265-unit leasehold boutique in Geylang (The Antares) produced a 100% win rate.

The D4 matched pair is the most controlled test in the data: same district, comparable unit sizes, completions within five years. The freehold mid-scale project outperformed both the smaller leasehold project and the larger leasehold mega. Tenure, not unit count, separated them.

The Martin Modern versus OUE Twin Peaks comparison adds the entry-timing dimension: same CCR D9 leasehold profile, similar scale, but different entry cohorts — and a gap of 75 percentage points in win rate.

What the data cannot cleanly resolve: whether a new boutique freehold RCR project bought today will replicate the outperformer pattern, or whether current entry prices represent a similar cycle-peak risk to what CCR buyers faced in 2012. The outperformers in this dataset — The Antares, Sky Everton, Forett@Bukit Timah — all entered at new-launch pricing before their respective market cycles peaked. That timing advantage is not reproducible on demand.

The clearest finding is about what did not cause the losses: project size. The riskiest condo, based on this evidence, was not the small one.


For more on how CCR resale outcomes diverge within the same region label, see The Same "CCR" Label Covered a $4.8 Million Gain and a $2.4 Million Loss. The Region Label Predicted Almost Nothing. and Singapore's Most Expensive Planning Area Is Also Its Worst for Condo Resale Profits.

Subscribe to Lucas Lim

Don’t miss out on the latest issues. Sign up now to get access to the library of members-only issues.
jamie@example.com
Subscribe