Useful calculated metrics for analysing Singapore’s property market.

We turn URA’s and HDB’s raw transaction records into district, market-segment, town, and project-level price trends, CAGR, and rental yield — private residential (condo and landed) and HDB resale, each figure graded by how much data actually backs it. Built for AI agents, available over REST and MCP, priced per call.

What you actually get back

Every computed figure ships with its own evidence, not just a number:

Example — GET /private/condo/districts/09/stats

{
  "district": "09",
  "property_category": "condo",
  "tenure": "ALL",
  "snapshot": {
    "median_psf": 2387.50, "p25_psf": 2240.00, "p75_psf": 2510.00,
    "transaction_count": 41
  },
  "cagr": {
    "value": 4.8,
    "unit": "percent_per_year",
    "confidence": "high",
    "inputs": { "transaction_count": 312, "years_used": 5.0 }
  }
}

$2,387median psf

4.8%CAGR (high confidence)

Condo and landed houses are kept separate throughout — their psf isn’t even measured the same way (floor area vs. land area), so blending them would be meaningless, not just imprecise. See Methodology for exactly how confidence is graded, including the cases where we deliberately return null rather than guess.

Access

REST

District, market-segment, and project stats, plus rental yield — each available as /private/condo/... or /private/landed/.... HDB resale data at /private/hdb/towns/{town}/stats (optional flat_type filter). Free listing endpoints (/private/condo/projects, /private/hdb/towns, etc.) help you find exact names first. Full spec at /openapi.json.

MCP

Streamable HTTP server at /mcp/ — the same data as agent-callable tools, with matching get_condo_* / get_landed_* / get_hdb_* tool names. A separate, permanently free mount at /mcp-free/ exposes just the discovery and sample tools, no payment handling at all.

Price	Account required	Settlement
$0.005 per call	No	USDC on Base, via x402

Want to try it free first? A fixed set of real examples is always free at /sample/districts/09/stats and the get_sample_data MCP tool - same response shape as the paid calls, no payment required.

Full machine-readable details: llms.txt · ai-catalog.json

Coverage

Private residential — district, market-segment, and project level, separately for condo and landed, derived from URA’s published transaction and rental data. HDB resale — town level, optionally split by flat type, derived from data.gov.sg/HDB records. No subdistrict geography for private property, no block/street geography for HDB. Recalculated monthly; check /meta before treating any figure as current.