At $4M, A Terrace House Has Returned Nearly Twice The Annual Rate Of A Premium Condo. There Is One Condition.

The conventional wisdom is correct: at $4M, a terrace house has beaten a premium condo. But the gap is bigger than most people realise. Terrace houses at Singapore's typical $4.22M entry have returned roughly 7% per year over a 4-year average hold. The best-performing premium condos at the same total investment delivered 3 to 3.5% per year. That is nearly double — and the difference compounds into real money over time. The condition: sell before the 2-year mark and landed's advantage collapses into a 1-in-4 loss rate. This article shows the numbers behind both sides of that trade-off.

The short version: At matched entry prices of around $4M, terrace houses have returned roughly twice the annualised rate of premium condos (≈7% p.a. vs 3–3.5% p.a.) over the past two years of data. But the 2-year hold threshold is not a footnote — it is the decision filter that changes the entire calculation.

Same capital, very different annual rate

Here is the comparison the entry price makes possible. Three premium condos at roughly $4–5M total investment, set against the typical terrace house at $4.22M.

PSF is not included here because landed PSF is calculated on land area while condo PSF uses strata floor area. Placing them in the same column would be misleading. All comparisons above are based on total transaction value.

The condo sellers at Aalto and Leedon held for 8 to 11 years and still came in below 3.5% per year. The terrace house sellers held for 4 years and nearly doubled that rate.

One important caveat: the condo holds in this table are longer — they span a more varied market cycle. If the premium condos had been measured over the same post-2020 window that drove the strongest landed gains, the gap might be narrower. The data cannot perfectly control for this, and the article cannot pretend otherwise. What it can say is that at matched capital and within the available dataset, the annualised return gap is real and substantial.

The hold period table is the spine of this article

Landed's edge is not unconditional. This table shows exactly where it breaks down — and where it becomes nearly unbeatable.

Source: Coalwood landed dataset, 2,004 terrace house transactions, Q3 2024–Q2 2026. Annualised return calculations use approximate midpoints for each hold band — treat as directional, not exact.

The jump that matters most is the first one: 76% to 95%, between under 2 years and the 2–4 year band. That single threshold — staying past the 2-year mark — eliminates roughly three-quarters of the loss risk. Stay past 4 years and losses become near-statistically irrelevant.

The annualised rate also moves across hold bands, and not in the direction you might expect:

• 4–7yr hold: ≈7.0% p.a. — the sweet spot for annualised efficiency
• 7–10yr hold: ≈5.6% p.a. — lower rate but 100% win rate, and average profit climbs to $2.47M
• 10+yr hold: ≈3.9% p.a. — approaching condo-level annualised returns, but $3.4M average absolute profit

The longer you hold, the more money you make in absolute terms — but the annual rate slows. Holding for 15 years at 3.9% p.a. starts to look similar to what Aalto achieved. The window where landed is genuinely in a different league, on both win rate and annualised efficiency, is the 2–7 year range.

Same district, completely different outcome: D15 terrace vs Aalto

Aalto sits on Meyer Road in District 15 — the same RCR city-fringe district as some of Singapore's best-performing terrace streets. So a same-district comparison is possible.

The holds are very different lengths, so this is not a clean apples-to-apples comparison. Aalto's sellers needed more than 11 years to book their profit. D15 terrace sellers needed just over 3 years to nearly double Aalto's absolute gain — and at a much higher entry price. The direction is clear even if the exact figures require some caution: landed in D15, at a higher capital commitment, has compounded faster and in a shorter window.

Aalto's 100% win rate is real and meaningful — none of its sellers lost money. But none of them approached the annualised return that the D15 landed cohort achieved in a fraction of the time.

The one place where both markets failed simultaneously

District 4 — Sentosa Cove and the broader Harbourfront/Keppel area — is the only losing district in the entire landed dataset. And it is also home to every major loss-making condo project in the data.

Landed D4: 12 transactions, average loss -$616,094, typical hold 1.1 years. The low transaction count means this is illustrative rather than conclusive — but the direction is consistent with what the condo data shows far more clearly.

Sentosa Cove condos:

The Residences At W Singapore Sentosa Cove (41 transactions, 0% win rate, 0.37yr avg hold) is excluded from this analysis — the near-zero hold period indicates almost entirely subsale activity, not resale performance.

Marina Collection sellers held for an average of 15 years and still lost $2.43M per unit. Seascape sellers held for 6 years and lost $936K. These are not short-hold mistakes. This is not what the hold-period table predicts should happen.

The distinction matters: for most of Singapore, holding past 4 years makes losses near-impossible for landed. Sentosa Cove condos show that when the purchase price was set at a 2007–2012 peak far above what the market could sustain, no amount of patience recovered it. Both the landed and condo data from D4 point to the same cause — not hold period, but entry price at a location-specific peak.

This is worth separating clearly from the broader condo comparison. The Sentosa losses are not evidence that condos underperform landed in general. The Trizon, Aalto, and Leedon — all held for 8–11 years — achieved positive returns at 95–100% win rates. The Sentosa projects show what happens when location pricing goes wrong on top of an already challenging asset class, and where the comparison to one district's outsized loss distorting everything is instructive.

What the data does and does not say

The data is not saying which asset class to buy. It is saying something more specific.

At matched entry prices of around $4M, terrace houses have historically returned nearly twice the annualised rate of premium condos — roughly 7% per year against 3 to 3.5% per year. That gap is large enough that it shows up clearly across multiple projects and thousands of transactions. It is not a rounding error.

But landed's annualised advantage is concentrated in the 2–7 year hold window. Under 2 years, the win rate drops to 76% — 1 in 4 sellers lost money, even in a period when the overall landed market was strong. The condos in this comparison — Aalto at 100% win rate after 11 years, The Trizon at 95% after 8.7 years — never approached that loss rate, even at lower annualised returns.

There is also a cycle-timing question the data cannot fully answer. The strongest landed gains in this dataset overlap with the post-2020 period, when landed properties in Singapore repriced sharply upward. Whether the next 4-year hold from today produces the same 7% per year is genuinely uncertain. The declining rate across longer hold bands — from 7% at 4–7 years down to 3.9% at 10+ years — suggests that some of the advantage is cycle-specific rather than purely structural.

What the data does resolve: the two-year threshold is not arbitrary. It appears consistently in the win-rate progression and represents the point where landed's loss risk drops from material to minor. A buyer who knows they need to exit within 18 months is looking at a very different risk profile from one who can commit to 4 years. That distinction is more important than the asset class decision itself.