HomeScope SG
A housing decision engine
for Singapore homebuyers
A buyer-first prototype that turns representative property scenarios into clearer housing decisions by layering lease, CPF, financing and MRT context into one comparison workflow.
Explore the prototypeA working decision-support prototype. Representative scenarios and imagery; not live listings.
Challenge
The harder problem is not finding homes.
It is knowing whether they still work for you.
Buyers already have discovery tools. The harder work is interpreting remaining lease, CPF implications, financing assumptions, MRT proximity and long-term fit across fragmented sources.
Mainstream property portals are primarily designed around discovery and enquiry flows. HomeScope explores a different layer: decision support, with the buyer’s constraints visible alongside the property.
- 01 →Search
Find possible homes
- 02 →Cross-check
Interpret the context
- 03 →Calculate
Understand assumptions
- 04 →Shortlist
Compare trade-offs
- 05 Decide
Verify before committing
Product Thesis
From property attributes
to buyer consequences.
Show the property in the buyer's time horizon
Purchase year and buyer age change how remaining lease should be interpreted.
Surface decision signals, not just raw attributes
Prioritise remaining lease, age-95 coverage, CPF signal, estimated mortgage and estimated cash top-up.
Keep the tool transparent and non-transactional
Representative scenarios and explicit assumptions—not live listings, duplicate entity-resolution or agent services.
Housing Decision Engine
The housing
decision engine.
Fig. 02 / From scenario to comparison
- 01
↓Property scenario
Price · property type · tenure · lease start · MRT distance
- 02
↓Buyer context
Purchase year · buyer age
- 03
↓Lease decision
Remaining lease at purchase · years to age 95 · covers buyer to 95? · lease remaining at 95
- 04
↓CPF signal
No lease-duration restriction / Eligible for CPF use, subject to limits / CPF use may be pro-rated / CPF use generally unavailable / Review
- 05
↓Financing assumptions
Loan type · LTV · rate · tenure · monthly CPF OA available
- 06
↓Estimated financing output
Estimated mortgage · estimated cash top-up · HDB tenure constraint / potential LTV prorating signal where applicable
- 07
↗Shortlist / compare
Bring the decision signals together, side-by-side. No overall winner or score.
HomeScope does not calculate personalised CPF entitlement, lender approval, MSR/TDSR outcomes or exact HDB loan approval. It uses transparent editable assumptions to help buyers interpret likely trade-offs.
Product Surfaces
Four surfaces.
One decision workflow.
Cards foreground remaining lease, age-95 coverage and estimated cash top-up instead of treating listings as static inventory.
A scenario expands into lease context, financing assumptions and CPF-related signals with explicit methodology.
A simplified MRT schematic supports station-led browsing and 500 m / 750 m / 1 km distance filtering.
Buyers can compare 2–3 shortlisted homes side-by-side on price, PSF, remaining lease, age-95 coverage, CPF signal, estimated mortgage, estimated cash top-up and MRT distance.
Current prototype captures · Source labels (PropertyGuru, 99.co, EdgeProp, SRX, HDB Resale, Ohmyhome and Stacked Homes) are illustrative within a representative dataset, not live feeds. Images are representative, not exact listing photos.
Key Product Decisions
The decisions behind
the decision support.
Purchase year became a first-class input
A property should be evaluated at the buyer's intended purchase point, not only in the present.
Age 95 became the more meaningful signal
The product asks whether the remaining lease covers the buyer to age 95, reflecting current CPF/HDB rules where age-95 lease coverage can affect CPF usage limits and HDB loan-to-value limits. This signal alone does not determine approval or exact CPF entitlement.
Financing assumptions are visible and editable
Expose loan type, LTV, interest rate, tenure and monthly CPF OA assumption rather than hiding them inside a calculator.
The prototype stays honest about scope
Representative dataset and imagery. “Open source site” redirects to source platforms. There is no live listing ingestion, duplicate entity-resolution or agent/contact workflow.
MRT is a decision filter, not decoration
The map was redesigned as a station-selection and distance-filtering interaction, with clearer selected-line labels and “Show homes” as the outcome. MRT Distance is proximity, not a routed walking-time calculation.
What Changed During Build
Building sharpened
the product thesis.
A better buyer product mainly needed cross-platform aggregation.
Aggregation alone did not create a meaningfully better decision.
Shifted the product thesis toward a housing decision-support engine using buyer context and consequences.
Generic lease-health scores would simplify the problem.
Arbitrary red/amber/green thresholds masked the actual Singapore housing consequence.
Replaced them with remaining lease, age-95 coverage and CPF-related signals.
Extra prototype affordances such as duplicate badges and contacted states made the product feel more complete.
Those UI states implied capabilities that were not implemented.
Removed duplicate-detection and agent-contact surfaces and tightened provenance/scope language.
Validation & Next Steps
A working prototype.
Not a validated outcome.
What the prototype demonstrates
- Buyer-specific lease interpretation using purchase year and age
- Age-95 coverage as a clearer decision signal
- Editable financing assumptions and estimated cash top-up
- MRT-led discovery with selectable distance radius
- Shortlist and 2–3 property comparison
- Transparent representative data and non-transactional scope
What still needs validation
- Whether these signals improve real buyer confidence or decisions
- How buyers interpret and trust CPF/financing signals
- Whether shortlist and comparison create meaningful repeat-use value
- Official / production-grade property and transaction data integrations
- Production-grade listing freshness and ingestion architecture
Evidence status / Prototype behaviour reviewed. No validated real-user outcome metrics are claimed.
The prototype tests one product thesis: property discovery becomes more useful when the constraints behind the decision are visible at the same time.
Explore the prototype