Why a 4.19% Yield Still Only Scores 64/100: Inside the QuantField Score's Five Components

Why a 4.19% Yield Still Only Scores 64/100: Inside the QuantField Score's Five Components

18 July 2026

---

Mandarin Gardens cleared our buy box yesterday at a 4.19% net yield — comfortably above the 3.2% floor — and still landed a QuantField Score of just 64/100. That is not a bug. A 99-year lease with 67 years left and a 2,670-unit supply wall from Grand Dunman, Emerald of Katong, and The Continuum landing next door are real, quantifiable drags, and the score is built to show them even when the yield math looks great. Every one of the five numbers behind that 64 is reproducible from public URA data and our own screen — there is no proprietary black box to take on faith.

---

1. Yield Strength (0-40) — the floor is 3.2%, not zero

Formula: min(40, max(0, (net_yield − 3.2) / 2.0 × 40)). A listing sitting exactly at the buy-box floor of 3.2% net yield scores 0 here — it cleared the screen, but only just. A listing at 5.2% or above (floor + 2.0 percentage points) scores the full 40.

Mandarin Gardens: 4.19% net yield. (4.19 − 3.2) = 0.99pp. 0.99 / 2.0 = 0.495. 0.495 × 40 = 19.8/40. It cleared the floor by less than a full point of yield, so it sits just under half marks on the single largest component in the score.

This is deliberate. Yield Strength is worth more points than the other four components combined, because net yield is the number that actually pays the mortgage. But scaling it against a 2.0pp band above the floor — rather than an open-ended scale — stops a merely-adequate yield from masquerading as an exceptional one.

2. Valuation Discount (0-20) — capped, not unlimited

Formula: clamp(10 − (pct_vs_median / 2), 0, 20), where pct_vs_median is the asking PSF's deviation from the URA district median transaction PSF for comparable units. At-median asking price scores 10. A listing 20% below median scores the full 20. A listing 20% above median scores 0.

Mandarin Gardens is asking 22.1% below the D15 median transaction PSF. Raw math: 10 − (−22.1/2) = 10 + 11.05 = 21.05 — but the formula caps at 20, so it scores the full 20/20. The extra 1.05 points implied by the raw number are simply discarded.

That cap is the point, not an oversight. A discount past −20% doesn't buy more score — it should instead prompt the question of why a unit is priced that far below its comps. In Mandarin Gardens' case, the answer shows up two components down: a lease past the 60-year mark and a supply glut next door. Steep discounts and structural risk tend to travel together in this market, and the score is built so cheapness alone can't paper over that.

3. Tenure Durability (0-15) — the 60-year cliff, priced

Freehold or 999-year leasehold: flat 15/15. For 99-year leasehold with R years remaining:

• R ≥ 60: 7.5 + (R − 60) / 39 × 7.5
• R < 60: 7.5 × (R / 60)²

The kink at 60 years is not arbitrary — it's the point at which CPF usage for a purchase starts getting pro-rated against the buyer's age, and where the pool of CPF-eligible resale buyers starts narrowing hard. Above 60, decay is linear and gentle. Below 60, the formula switches to a squared term, so the same number of years lost costs disproportionately more score.

Mandarin Gardens has 67 years remaining — just above the cliff. 7.5 + (67−60)/39 × 7.5 = 7.5 + 1.35 = 8.85/15. Not full marks (it's still decaying toward the cliff), but nowhere near the penalty zone.

4. Data Confidence (0-10) — how much of this can we actually verify

Formula: min(n/10, 6) + (4 if rent is live URA data, else 0), where n is the depth of the URA transaction comps sample. A thin comps sample caps out low on the first half; a hardcoded rent estimate (rather than a live URA rental contract) forfeits the second half entirely, regardless of comps depth.

D15 is one of the most heavily transacted resale districts in Singapore, so Mandarin Gardens' comps sample clears the depth cap, and its rent input is drawn from live rental contracts rather than an estimate: 10/10.

5. Supply Risk (0-15, inverse) — more competing stock, lower score

Formula: 15 × max(0, 1 − pipeline_units / 4000). Zero competing units in the pipeline scores the full 15; a pipeline approaching 4,000 units scores toward zero.

D15's current pipeline includes roughly 2,670 units still to complete across Grand Dunman, Emerald of Katong, and The Continuum. 15 × (1 − 2670/4000) = 15 × 0.3325 = 5.0/15. That's the second-weakest component in the whole breakdown, and it's weak for a concrete reason: three large new launches are about to compete for the same rental and resale demand pool Mandarin Gardens draws on.

---

The Full Worked Example: Mandarin Gardens at 64/100

Read the table left to right and the BUY call makes sense without a single hedge: the yield clears the floor, the price is genuinely cheap against comps, but two real structural facts — a lease past its prime and a nearby supply wall — hold the score in the mid-60s rather than the 80s. That's not a contradiction. That's what a transparent score is supposed to do: let a good deal be a good deal and show you exactly what you're trading off to get it.

Why Hillcrest Arcadia Scores Differently on the Same Five Inputs

Hillcrest Arcadia cleared our buy box on 13 July 2026 at a 3.76% net yield — also above the floor, but with a thinner data foundation. Run the same three components through the identical formulas:

A 14.4-point gap opens up on just three of the five components, before Valuation Discount or Supply Risk enter the picture at all. Hillcrest Arcadia's yield clears the floor by less margin (0.56pp vs 0.99pp), its 53-year remaining lease sits on the steep side of the 60-year cliff rather than just above it — 53² vs 60² inside that squared term costs it real points that 67 years never touches — and its 32-deep rent comps sample means the Data Confidence formula can't award the full comps-depth marks no matter how good the rest of the picture looks.

Both properties cleared the same 3.2% buy-box floor. The score is what tells you they are not the same bet.

---

What This Means For You

Buyers: don't read the headline score as a single verdict — read the five numbers behind it. A 64 driven by supply risk and lease decay (Mandarin Gardens) is a different risk profile from a 64 driven by thin rental data, even if the totals matched. Ask which components are dragging the score before you ask what the total is.

Sellers: if you're listing a 99-year leasehold property, know that the Tenure Durability formula punishes years lost below 60 roughly twice as hard, via the squared term, as years lost above it. A unit at 61 years and a unit at 53 years are not "8 years apart" in score terms — they're on different sides of a cliff.

Agents: every component here is reproducible from a client's own URA comps pull and the listing's TOP date. When a buyer asks why a deal scored what it did, you can rebuild the math on the spot instead of citing a proprietary index they have to take on faith.

---

QuantField Score is a screening tool, not investment advice. Scores are calculated from URA transaction and rental data plus listing inputs at time of screening and can change as new comps and supply data are published. Always verify tenure, comps, and pipeline figures independently before transacting.