South Florida Price-per-SqFt & Market Performance Index

Two separated layers

This index combines two datasets kept explicitly separate and individually labeled throughout. No MLS figure is ever presented as deed-derived, and no deed figure as MLS-derived.

Layer A — FDOR public deed transactions

The Florida Department of Revenue (FDOR) NAL assessment roll — the annual county record of every parcel, its qualified sale prices and years, deed-qualification codes, just (assessed) value and living area. Residential parcels only (DOR use codes 1, 2, 4, 5) across Miami-Dade, Broward and Palm Beach. Only qualified arm's-length sales are counted (the roll's market-sale flag, i.e. FDOR deed-examination codes such as 01/02), with nominal transfers removed where sale_price / just_value falls outside 0.2–5.0 (dropping $100 quitclaims and intra-family transfers). Each sale year is drawn from its own assessment roll so the just-value screen is contemporaneous, avoiding a bias against older sales; the prior year is pooled across its two contemporaneous rolls and deduplicated by parcel so county-level roll-loading gaps do not distort year-over-year coverage.

Layer B — MLS current-market context

Closed-sale liquidity, days-on-market and cash share come from MLS closed-record rollups (trailing 12 months). These describe the current market and appear only as context alongside the deed statistics — never blended into them. Every MLS figure on this page is labeled MLS.

Headline statistic & dispersion

The published headline is the median qualified transaction $/SqFt = sale_price / living_area per market/building. Mean, quartiles (Q1/Q3), standard deviation and coefficient of variation are computed and retained; dispersion is published where it is itself the finding. Markets below 40 qualified sales, and buildings below 8, are suppressed so thin samples never enter the rankings.

Movement, not appreciation

The "YoY $/SqFt Δ" column is the year-over-year change in median transaction $/SqFt — a compositional measure of what sold, not a constant-quality appreciation rate. We tested a true repeat-sales appreciation measure (the same property transacting in two periods, both qualified arm's-length) against the deed roll and found it insufficient to publish: the roll's two-slot sale history yields only ~1,600 tri-county qualified pairs with a ~1-year average hold — a flip-dominated, mix-biased sample too thin to break out by market. We therefore report the median-movement measure and label it honestly rather than overstate a repeat-sales index the source cannot yet support.

Geographic & building attribution

Parcels attribute to a market by their physical-situs city (phy_city) matched to our geography; 87.3% of qualified transactions matched a known market (the remainder — largely unincorporated or out-of-scope situs strings — are excluded, not reassigned). Buildings attribute by a leading-folio match to a curated register of 224 named towers; building coverage is therefore a subset of major buildings, disclosed as such.

On differentiation

County deed rolls are public records — anyone can obtain them, and we claim no exclusive access. What distinguishes this index is the analysis: correct arm's-length qualification and contemporaneous nominal-transfer screening, per-vintage roll selection, building-level granularity below the ZIP, a sale-to-just-value view that only a deed-plus-assessment join produces, and an explicit separation of public-record history from current MLS behavior — computed consistently and disclosed openly.

Update cadence & vintage

Deed-derived figures refresh with each annual FDOR NAL roll; MLS context refreshes nightly. This edition reports 2024 qualified sales (2025 FDOR roll). A machine-readable companion of every published figure is available at https://www.andresjara.com/research/price-per-sqft-index/data.json.