HypeCity collapses 19 underlying variables — yield, price-to-median, climate exposure, political stability, walkability, safety, and 13 more — into a single Investment Signal label per persona. The labels are deliberately blunt. A sophisticated investor can drill into the underlying scores; a busy investor just needs to know whether the deal is worth a closer look.
Strong Fit
High persona alignment AND high data confidence (CI < 2 points wide).
The neighborhood scores in the top quartile against the chosen persona's weighted criteria — yield, lifestyle index, climate resilience, or whatever the persona ranks first — and the underlying data is dense enough that the confidence interval is tight. This is our highest-conviction call.
Fits at Right Price
Persona alignment is solid, but the listing price sits more than 25% above the neighborhood median.
The fundamentals are there; the listing isn't. This signal is common in heated submarkets where a strong neighborhood gets paired with an aspirational asking price. We surface it so buyers can negotiate from a position of clarity, not walk away from a thesis that's actually intact.
Does Not Fit
Low persona alignment across the heaviest-weighted dimensions for the chosen persona.
For a FIRE persona that's a thin yield combined with a stretched price-to-rent ratio. For a Lifestyle First persona it's poor walkability combined with weak neighborhood amenity density. The signal is persona-specific — a Does Not Fit for one buyer is often a Strong Fit for another.
Low Confidence
Data confidence below 30% or CI more than 4 points wide.
When the inputs aren't dense enough to defend a verdict, we say so. Rather than print a number we wouldn't bet on, we flag the analysis and the UI tells the user "Low confidence — verify with current comps." Honesty is the moat.
Why four, not five or ten? Four signals map cleanly to four actions: pursue, negotiate, pass, or research further. Anything more granular becomes false precision — a noise-fitter, not a decision tool.
§2 — Persona framework
Same neighborhood, different verdicts.
The same neighborhood can be a Strong Fit for one investor and a Does Not Fit for another. That isn't a contradiction — it's the entire point. The signal is meaningless without knowing who the investor is.
Every analysis runs through one of 6 personas the user picks before submitting a listing. Each persona ships with its own weighted criteria — what it cares about, what it tolerates, what it treats as a deal-killer. We're growing the persona library toward 30+ as the catalog scales internationally.
Worked example · Palermo, Buenos Aires
Lifestyle First
Strong Fit. High walkability (94/100), dense café and gallery layer, established creative-class demographics. Currency volatility is a wash because the persona doesn't weight yield at all; the use case is residency, not cash flow.
FIRE
Fits at Right Price. The yield math is plausible in pesos but the dollar exit risk is meaningful, and the listing's USD-equivalent ask sits 30% above neighborhood median. Negotiate to median, the FIRE thesis works; pay ask, it doesn't.
Diversifier
Strong Fit. The persona explicitly seeks USD-decorrelated exposure with low jurisdictional overlap to existing US-heavy portfolios. Argentina's political risk gets re-weighted as a feature, not a bug.
This is moat #1 of 4. Most other real-estate scoring tools return one verdict per listing — the same verdict for every buyer. We return a verdict shaped by who's asking. That difference compounds with every persona we ship.
§3 — Two-layer architecture
Deterministic floor, AI ceiling.
The signal pipeline is intentionally two-layered. Layer 1 is rule-based and auditable; Layer 2 is AI-synthesized and contextual. The split exists because some calls should never be subjective, and others can't be made by rules alone.
Layer 1 · Deterministic
Auto-flags — non-negotiable thresholds.
A small set of hard rules run first. They produce automatic flags before the AI ever sees the data, and they cannot be overridden by Layer 2:
Catalog signal — is this neighborhood in our covered set with fresh data (refreshed in last 90 days)? If not, the analysis routes straight to Low Confidence.
Price vs. neighborhood median — list price more than 25% above local median triggers a Fits at Right Price flag regardless of other fit signals.
Climate risk threshold — properties in 100-year flood zones or with NOAA urban-heat-island delta above +4°C trigger a deterministic risk note that survives into the final label.
Safety floor — neighborhoods below the safety percentile floor for the chosen persona auto-route to Does Not Fit, no matter how strong the yield or walkability scores.
Layer 2 · AI synthesis
Five-factor verdict — what the rules can't see.
Once the deterministic flags are set, the AI layer synthesizes the verdict across five weighted factors. Each factor is computed from the underlying 19 variables and the persona weights:
Yield — rental income relative to price, normalized for local cost-of-living and tax drag.
Price — list-vs-median, persona-adjusted appreciation potential, sensitivity to a 10% drawdown.
Climate — composite of flood, heat, wildfire, and grid-fragility signals on a 30-year horizon.
Political stability — country-level governance, rule of law, foreign-ownership protection, capital-controls history.
Why split it? A pure-rules system can't read a market the way a generalist can; a pure-AI system can quietly drift past a hard safety floor. Layer 1 guarantees the floor. Layer 2 reads the room.
§4 — Confidence intervals
Confidence is honesty, not weakness.
Real-estate data is noisy. Even in the best-covered jurisdictions, comparable sales arrive late, MLS feeds disagree, and rental data is a moving target. We refuse to print a verdict more confident than the data supports. Every signal ships with a confidence interval (CI), expressed in score points.
CI < 2
High confidence. Dense source coverage, multiple cross-checking feeds, recent refresh. The verdict is the verdict. Drift is unlikely to flip the label between today and your closing date.
CI 2–4
Medium confidence. Direction is reliable, precision is approximate. Treat the signal as orientation rather than a final word — the underlying drivers are likely real, but a single noisy comp could shift the magnitude.
CI > 4
Low confidence — verify with current comps. The UI surfaces this banner directly on the result page. We're telling you the inputs are sparse enough that we can't defend either a Strong Fit or a Does Not Fit. Use the analysis as a starting hypothesis and pair with a local broker's current-comp set before committing capital.
The strategic point. Most real-estate tools surface a single confident number. We surface a range, and we tell you when the range is too wide to act on. That isn't a UX weakness — it's the entire trust contract. A signal you can't trust isn't worth shipping.
§5 — SEC & compliance
Why we say "FIT" not "BUY".
The labels HypeCity surfaces are deliberately non-prescriptive. Strong Fit is a statement about the relationship between a neighborhood and a persona — not a recommendation to transact. We never label anything BUY, SELL, HOLD, or any other word that resembles a securities-style recommendation.
That language is intentional and load-bearing. HypeCity is an information platform. We are not a broker-dealer, not a registered investment adviser, and not a fiduciary. Our outputs are research signals you weigh against your own goals, capital position, and risk tolerance.
For US tax residents transacting cross-border, the diligence layer goes further than the signal:
FATCA — Foreign Account Tax Compliance Act filings may apply when ownership routes through non-US entities. We surface a flag on international analyses but do not file on your behalf.
FBAR — FinCEN 114 reporting requirements for foreign-account interests above the threshold are the user's responsibility. Talk to a cross-border tax professional before wiring funds offshore.
FIRPTA — Foreign Investment in Real Property Tax Act considerations for non-US persons disposing of US real estate flow into our US-listing analyses as a disclosure note, not as a scoring penalty.
Not investment advice. Every analysis HypeCity produces is research and information only. Nothing on this site constitutes a recommendation to buy, sell, hold, or finance any property or security. Consult a licensed advisor in your jurisdiction — and a licensed broker in the jurisdiction where you transact — before acting on any signal.
§6 — Sources
Where the numbers come from.
Every variable behind the Investment Signal is sourced from a public dataset or a fair-use research aggregate. We do not source from any single private data partner whose terms would prevent us from showing our work.
US Census Bureau — Tract-level dataDemographic shift, population trend, income mix, education attainment. Anchors the persona-fit math for every US neighborhood.
Redfin Data Center — Median price aggregatesNeighborhood-level median list and sale price, days-on-market, listing velocity. Public CSV downloads, refreshed monthly.
Walk ScoreWalkability index per geocoded coordinate. Pairs with the EPA Smart Location Database's intersection-density signal for street-grid honesty.
OpenStreetMap — POI densityCafé, restaurant, gallery, grocery, and transit-stop density per neighborhood. Open data license; the lifestyle layer is built on it.
Public yield data — RentCafe, RentometerRental rate aggregates by neighborhood and property type. Used as the yield numerator in the gross-yield calculation.
Climate Central + FEMA flood mapsSea-level-rise projections, flood-zone classifications, urban heat-island deltas. The Layer 1 climate floor is built on these.
Zillow public ZHVIZillow Home Value Index — neighborhood and ZIP-level value indices used as a triangulating signal against Redfin medians. Public research download, used under fair use.
For the full source list across all 19 Premium variables (including SBA loan originations, EPA Envirofacts, NOAA Urban Heat Island, FBI UCR, and FDIC bank-branch density) see the methodology overview.
§7 — Scoring flow
How a listing becomes a signal.
The pipeline runs in five stages. Raw data on the left, a signal label on the right. Persona weights enter immediately after ingest, deterministic flags fire before the AI layer, and confidence is computed in parallel against the final label.
Stage 1
Raw data (19 vars)
→
Stage 2
Persona weights applied
→
Stage 3 · L1
Deterministic flags fire
→
Stage 4 · L2
AI synthesis (5 factors)
→
Stage 5
Signal label + CI
Layer 1 · Deterministic
Layer 2 · AI
Final output
Confidence intervals are computed alongside Stage 4 from source density, recency, and cross-source agreement. A signal that clears Stage 4 but has CI > 4 is downgraded to Low Confidence at Stage 5, regardless of how clean the verdict looked.
§8 — Try it
See the signal on a real listing.
The fastest way to understand the methodology is to run a listing through it. The free tier returns the full Investment Signal — persona-weighted, CI-flagged, deterministic-checked.