Why two identical apartments price differently

Two units with the same floor plan in the same tower routinely transact at meaningfully different prices. Floor, view, orientation, condition and timing all move the number, and their effects stack. This is why tower-level or district-level averages are a weak guide to what a specific unit is worth: an average describes the pool, not the unit, and the spread inside the pool is where deals are actually won and lost.

Start inside the tower. Higher floors typically command a premium, but not linearly, and not identically in every building. View is priced ruthlessly: the same plan facing water and facing a service road are different products. Orientation drives light and heat, which tenants feel and pay for. Condition and fit-out separate renovated from original stock. And position within the floor plate matters: distance from lifts, exposure to corridor noise, corner versus middle.

Then add time. Two identical units sold six months apart sit in different markets. In a fast cycle, the timing difference alone can exceed every physical premium combined. A comparable is evidence about the market on its transaction date, not about the market today, and it decays.

An average price per square metre for a tower or district compresses all of that spread into one number. It mixes high floors with low, renovated with tired, water views with road views, last month’s sales with last year’s. The result is a number that describes no actual unit. Used as a pricing anchor, it systematically overprices the weak units and underprices the strong ones, which is precisely how mispriced listings are born, and how they get bought by whoever measures the spread instead of the average.

Illustrative, deliberately abstract, no real project. Take two units with an identical plan in one tower. Unit one: high floor, open view, corner position, original condition. Unit two: low floor, facing the neighbouring building, mid-corridor, freshly renovated. A district average prices them identically. A unit-level read prices the view and height premium on one side against the renovation premium on the other, weighs each against what unit-level comparable evidence actually shows for this building, and lands on two different numbers, both defensible, neither equal to the average. The average was not wrong; it was answering a different question.

Pricing at unit level means measuring the premium each attribute earns in this building and this pocket of the market, from transaction evidence, rather than applying rules of thumb. It also means selecting comparables like a practitioner: same demand pocket, similar attributes, recent dates, with adjustments that are explicit rather than intuited. Every adjustment is a claim, and in a glass-box system each claim is traceable to the evidence that supports it.

How big can the spread inside one tower be?

Large enough to decide deals, and it varies by building and market phase, which is why no universal figure is honest. The practical point is not the size of the spread but its existence: if you price off the average, someone pricing off the unit is ahead of you in both directions.

Is a renovated low-floor unit better value than an original high-floor unit?

There is no general answer; that is the point. Each premium has a measurable market value in a specific building, and the right comparison prices both packages against unit-level evidence. The buyer who wins is the one who knows which premiums the local market actually pays for.

Do these unit-level differences matter for rental yield too?

Yes, and sometimes differently than for sale prices. Tenants pay for light, view and condition, but they price them on their own scale. A unit can carry a sale premium its rental market does not fully reimburse, which shows up directly in yield. Pricing purchase and rent from the same average hides exactly this.

Related reading: How pricing models see a building, and Glass-box AI: why every output traces back to a source.