True medians of actual closed sales, now split by property type. Each panel answers compared to what? three ways: its own decade, a year ago, and its peers side by side. Click a district, then a subdistrict, down to the real sales underneath.
True medians. Every price, sale-to-list, and price-per-square-foot figure is the median of individual closed sales (POTM Command LISTINGS_SOLD, 42,547 closings used), computed live for each property type, geography, and rolling-12-month window from May 2016 to present. Half of homes sold above the number, half below. No blending across segments.
Property types follow the project groupings: Single Family (Single Family Residence), Condo / Townhouse (Condominium plus Townhouse), Tenancy in Common, and MultiPlex (Duplex, Triplex, Quadruplex, 5 or more units). Thin types in small districts will read Thin or Anecdotal; the chip says so.
Price per square foot controls for size, the honest partner to a median. REVIEW-flagged sales with missing square footage are held out of this metric only (their price-per-sqft would be zero) and remain in price, sale-to-list, and Days on Market.
What moved the number buckets sales by list price, not sale price, so a tier label stays fixed to what the seller asked and the sale-to-list percentage beside it reads cleanly: it is how that list-price tier bid versus its own asking, in the current window. The bars show how each tier's share of sales shifted versus a year ago.
Agent read. The interpretive commentary is opinion, labeled as such and kept separate from the data. It is generated from the patterns above plus the macro backdrop (your governed FRED and BLS tabs) and current local reporting, and it carries both seller and buyer implications. It is a read on present conditions, not an appreciation forecast, and not legal, tax, lending, or investment advice. Local news and on-the-ground field notes are refreshed each cycle.
Macro and financial backdrop. Rates, inflation, jobs, and the US median home price come from your governed FRED and BLS archives, each shown against its level a year ago. Equity indices and AI capital are external, indicative, and dated, drawn from public reporting and labeled as such until ingested into POTM Command. The house-premium signal is computed from this tool's own true medians. AI and tech capital citations are held to a rigid three-month window; any source older than the cutoff shown on the note is rejected at build time.
Governance. Readiness gate READY_FOR_REPORTS confirmed 2026-06-11. Source-deduplicated by listing number. 5 sales with sale-to-list over 200 percent excluded as entry errors. Quarantined rows excluded by source. Reliability chips on POTM Command thresholds (district rolling-12: 20 / 10 / 5; subdistrict: 15 / 10 / 5).
Data deemed reliable but not guaranteed, subject to change, correction, and revision. No claim of 100 percent accuracy. Prototype format for review.
The Agent Lens · part of Pulse on Today's Market · data via POTM Command governed analytics.
Paulo Serna · Level Up Group, Compass · paulo@levelupgroup.com · 408-834-9161
Want this as a one-page read to share or print?
San Francisco residential market intelligence, built on real closed sales and refreshed each cycle. Paulo Serna · Level Up Group at Compass.
