Rubric v3 · published
How a project gets its score
This page is the whole method. If a number on this site cannot be traced back to something here, it should not be there.
Two rules
A score is a judgment that shows its work
Seven categories, each with a published rule. An analyst model reads the complete evidence file and scores every category against that rule, writing its reason and citing the sources it used. Code enforces the guardrails below. The rules are stored versioned in the database; a score cannot change unless the evidence or the rule does.
A fact needs a source
Every published field carries a ledger entry: the value, where it came from, when it was retrieved, how confident that retrieval was. Each project page shows its source count and links the ledger. What we could not verify is marked unverified and never counts toward a score.
Where facts come from
Ranked. A fact is taken from the highest tier that carries it.
- First-party The developer's own website, offering materials and condominium documents.
- Public record Miami-Dade deeds, mortgages, lis pendens and liens; building permits and inspections; Florida DBPR condominium filings; Sunbiz corporate records; FEMA flood maps; the Property Appraiser.
- Press The Real Deal, Bisnow, South Florida Business Journal, PROFILEmiami, The Next Miami.
Broker sites, listing aggregators and ranking blogs are used only to discover that a project exists. Nothing they publish is treated as a fact or reproduced here.
The seven categories
Market momentum Is the market validating this project? ×1.3
Sell-through pace relative to launch and unit count; price movement since launch; quality of capital (JV partners, recorded construction loan, named lender); brand and design pedigree. Calibration: >70% sold with a recorded institutional loan = 9–10. >50% sold or a recorded loan = 8. Launched with a brand and sponsor but no pace evidence = 7. Stalled, relaunched, or lender trouble = ≤4.
Inputs sales_pct_soldlaunch_datetotal_unitsconstruction_loan_centsconstruction_lenderdeveloper.partnersbrandquotes
Delivery Will it be built, on the timeline claimed? ×1.2
Construction stage, recorded financing, sell-through, sponsor delay record, open project-scope litigation. Calibration: topped off with recorded financing and >70% sold = 9.5. Under construction with financing = 8. Under construction without a recorded loan = 6. Planned or pre-construction with no financing = ≤4. Open project-scope litigation caps this at 6 (code-enforced).
Inputs statusconstruction_loan_centssales_pct_soldcompletion_eta_textlegal(scope=project)developer.avg_delay_months
Developer Has this sponsor delivered, and what happened afterwards? ×1
Delivered record, years operating, capital partners, delay record, cited defect suits per delivered building, cited lender defaults. Calibration for this market: the established Miami condo houses (Related, Swire, Terra, PMG, Fortune International, CMC and peers) are 8–9 unless a cited default or a pattern of association suits pulls them down; an institutional sponsor with a named capital partner and drawn senior debt (e.g. a Starwood JV) is 7–8 even without a Miami-Dade condo TCO on our file; a genuinely unproven or thinly capitalised sponsor is ≤5. Our delivered count only covers 10+-storey Miami-Dade condos with a TCO — a null there is our gap, not a mark against the sponsor.
Inputs developer.projects_delivereddeveloper.founded_yeardeveloper.defect_suitsdeveloper.lender_defaultsdeveloper.avg_delay_monthsdeveloper.partnerslegal(scope=developer)
Location and view Will the thing you are paying for still be there in five years? ×1
Site quality and view durability. Calibration: Brickell, Miami Beach, Bal Harbour, Fisher Island, Coconut Grove, Edgewater bayfront and Downtown waterfront are prime by definition — baseline 7. Direct ocean or bay frontage with no buildable parcel in the view = 9–10. An interior prime site with taller permitted neighbours = 7, not 5. Deduct only for a cited, specific threat to the view or the site (a permitted tower in the sightline, a flood finding), and say which.
Inputs addressneighborhoodnearby_developmentwalk_scoretransit_notefema_zone
Value Is this priced with or against its peers? ×1
Price positioning against the neighbourhood band and named comparables, adjusted for brand, spec and frontage. The market's own verdict counts: a building >70% sold at its asking price is being paid that price — score 7+ regardless of whether we hold a band. In band for a superior product = 8. Well above band AND slow sell-through = ≤4. Never default to 5 because comparables are absent.
Inputs ppsf_avg_centsprice_from_centsprice_to_centsunit_sqft_fromunit_sqft_toneighborhood.ppsf_bandamenities
Liquidity How easily could you sell or lease it again? ×0.8
Resale and leasing case: unit count and size mix, brand recognition, frontage, depth of the neighbourhood resale market, rental and short-term policy. Calibration: branded or low-unit-count building in a prime neighbourhood = 8. An unpublished rental policy is a flag on the label and NOT a deduction; deduct only for a cited restrictive policy or a thin resale market.
Inputs str_policytotal_unitsunit_sqft_frombrandneighborhood
Buyer protection If the project stalls, what happens to your money? ×0.7
Deposit exposure, stage-adjusted. Florida escrows the first 10% by statute; terms above that are a disclosure, not a statute. Calibration: topped off with recorded financing and >70% sold = 8–9 (the money is nearly home whatever the escrow terms); under construction with financing = 7; planned with no financing = ≤4. Unpublished escrow terms are a FLAG in the reason and carry NO deduction. Published escrow above 10% adds; open filings naming the project subtract.
Inputs deposit_structuredeposit_escrow_pctstatussales_pct_soldlegal(scope=project)
The overall score is the weighted mean of the scored categories, on a 0–10 scale. A category the analyst cannot score is left out of the mean rather than filled with a midpoint.
Three rules that are not left to judgment
Nothing above 7 without a source. A category may only score higher than 7.0 when the analyst cites at least one URL it relied on. Enthusiasm is not evidence.
Filings cap delivery. Any open civil action or lien that names the project caps its delivery score at 6.0 until it is closed, dismissed or settled. Matters concerning the developer elsewhere inform the developer category only.
Florida escrows only the first 10%. Deposit terms above that are a disclosure, not a statute. Where they are not published we say so on the page; the buyer-protection score is judged on the stage of the building and the sell-through, which is what actually determines how exposed a deposit is.
Questions
- Who decides the scores?
- An analyst model reads the complete evidence file for a project — every sourced fact, the developer's record, court filings, nearby permits, the narrative report — and scores each category against the rule published above, writing its reason and listing the sources it relied on. Code then enforces the guardrails: no category above 7 without cited evidence, open filings naming the project cap delivery at 6, and fewer than four scored categories means no overall score. Earlier versions were pure arithmetic over a handful of fields; that produced confident-looking numbers from our own research gaps, which is why this changed.
- Can a person change a score?
- Yes, and you will see it. An editor may apply an adjustment to the overall score with a written reason; the analyst score, the adjustment and the reason are all printed on the project page. The same rule applies to every project, including the one our founder owns in.
- What counts as a source?
- In order: the developer's own website and offering materials; Miami-Dade County public records — deeds, mortgages, permits, court filings; Florida DBPR condominium filings; then reputable press. Broker sites and listing aggregators are used only to discover that a project exists. Nothing they publish is treated as a fact here.
- Why do some figures say 'Not published'?
- Because they are not published. Where a developer has not disclosed escrow terms or association dues, we say so in the reason and flag it, rather than estimating a figure. A gap is scored from what is known around it — stage, sell-through, financing — not parked at a midpoint, and a category with nothing to judge from drops out of the average instead of dragging it.
- What does 'provisional' mean?
- Fewer than five of the seven categories have cited evidence behind them. The score stands, but read it as a first pass; the page shows how many categories are evidenced and what is missing from each.
- Why do two sources disagree on the same page?
- Because they do, and hiding it would be the dishonest choice. Where sources conflict we publish the higher-tier value and show the alternative beside it with its source, so you can judge.
- Why is litigation split into two lists?
- A lawsuit against a developer about a different building is real context for their track record, but it is not a mark against this project. Only matters naming the project affect its delivery-risk score. Both are shown, clearly separated.
- How often does this update?
- Standing monitors watch each published project daily for price changes, construction milestones, financing and filings. County records are swept monthly. A recomputed score is applied automatically only when every category moves less than 0.3 and every changed input is verified; anything larger waits for a person.
- Is this investment advice?
- No. It is a structured reading of public information, and it can be wrong or out of date. Verify anything material with the developer and your own attorney before you rely on it.