The 2.3× rule: developers price your condo years before it exists

Every launch weekend, buyers debate whether the price is fair — as if it were decided that week. It wasn't. It was set two to three years earlier, at a state land auction most buyers never read about, by a formula so stable we can publish it. Here's the rule, the evidence across 44 launches, and the forecast for every major site in the pipeline.

Here's a trade secret that shouldn't be one. When a developer wins a Government Land Sales site, the launch price of the condo that will eventually stand on it is already, for practical purposes, decided. Not to the dollar — but to a band tight enough to plan a purchase around. The reason is structural: every bidder at that auction worked backwards from the same inputs. Construction cost per square foot. Financing over a three-to-four-year build. Marketing, agency commissions, the developer's required margin — and, since the 2023 harmonisation of ABSD deadlines, a hard five-year clock to sell everything or hand the state a punitive tax. Feed the same costs and the same clock into every bidder's spreadsheet, and the bid that wins implies the launch price that must follow. The auction result is a forecast. Almost nobody reads it that way.

We do. Our model tracks every major launch since 2022 against the land price its developer paid, per square foot of buildable area. The ratio between the two — launch PSF divided by land PSF — is what we call the multiplier, and it is remarkably disciplined:

Launch avg PSF ÷ land PSF (per plot ratio), 99-year GLS sites, launches 2022–2026. Band calibrated by leave-one-out backtest, not standard deviation.

Notice the gradient: suburbs 2.33, city fringe 2.12, core central 1.95. This isn't arbitrary — it's arithmetic. Construction cost doesn't care where the tower stands; a square foot of concrete costs roughly the same in Tengah as in Newton. When land is cheap, that fixed cost stack is a huge share of the final price, so the ratio of launch price to land price stretches. When land is $1,800 psf, the same cost stack is proportionally small, and the ratio compresses toward the developer's margin. Understand that one sentence and you can sanity-check any launch price in Singapore with a phone calculator.

A selection from the model — chosen to show the rule holding across regions, years and market moods:

Land PSF per plot ratio; launch PSF = launch-period average; 99-year GLS sites. Full 44-launch dataset lives in the POV map's Analysis tab.

Two honest observations from the full dataset, including the parts that don't flatter the rule. First: the multiplier has been drifting upward since 2022. Lentor Modern launched at 1.75× its land cost; by 2025–26, suburban launches routinely clear 2.4–2.7×. Same corridors, fatter ratios. That drift is the launch market's real price inflation, and it's been running hotter than the resale index — developers have been passing through cost increases and widening the premium, because buyers kept absorbing it. Watch the Lentor rows: four launches in one corridor, land prices falling from $1,204 to $985 as the state released supply, launch prices rising anyway. That's the multiplier doing the work.

Second: the multiplier predicts the price, not the party. Tengah at 2.58× sold 99% on day one; Lentor Hills at a modest 1.96× took years to sell down. Buyers on launch weekend respond to absolute quantum and the location's story, not to the ratio — which is exactly why the ratio survives. Nobody at the showflat is checking the developer's homework. You should be: it tells you, years early, what's coming.

This is the practical payoff. Apply the region multipliers to land already sold, and the next two years of launch pricing is sitting in public data. Estimates below use the calibrated band (roughly ±15–17%); they assume today's market conditions persist, which is the model's biggest assumption and the reason these are bands, not points:

Implied = land PSF × region multiplier; band from leave-one-out calibration (thin buckets widened). Indicative, not a valuation. Full pipeline in the POV map.

Sit with the highlighted rows for a moment. The model says Hougang — heartland Hougang — is on course for a launch around $2,750 psf, and Bedok around $3,100. If those numbers feel impossible, remember they felt impossible in Tampines too, until Parktown printed $2,360 and sold 87% in a weekend. This is the single most under-appreciated fact in the current market: the state's own land sales have already locked tomorrow's suburban launch prices at levels 40–70% above today's resale stock next door.

If you own near an upcoming site: the launch two years out is your exit's marketing campaign, paid for by someone else. Every launch in our dataset reset its neighbourhood's price conversation upward; owners who sold into the months around a major launch consistently faced buyers who'd just been educated — by a showflat — that the "cheap" option was the resale. Know your local land deals and time accordingly.

If you're buying resale: invert it. Find the corridors where land has transacted at prices implying launches far above current resale — Hougang and Bedok top that list today — and understand that the gap between what you pay now and what the showflat will ask in 2027 is your margin of safety. This is the closest thing to buying with the answers in advance that this market offers.

If you're eyeing a new launch: compute the multiplier before you queue. A launch pricing at or below its region's average ratio is pricing to move; one far above it is asking you to fund an above-market land bid or an above-market margin. Ask which, and make the developer's homework part of yours. (Freehold en-bloc sites run a separate, thinner book — collective-sale launches average 1.99×, because the land came dearer and the tenure premium gets baked in.)