Frequently Asked Questions

HomeBidder uses a comparable-sales (comp) approach as its primary method. It finds recently sold properties near the subject with similar characteristics, then adjusts each comp's sale price for differences in lot size and square footage using a log-linear hedonic model. The adjusted comp prices are averaged to produce a base fair value, which is then modified by any applicable discounts (e.g., TIC ownership, fixer condition). When fewer than three comps are available, the model falls back to a price-per-square-foot estimate derived from the same comp pool; when no comps exist at all, the list price is used as a proxy.

Comps are drawn from recent MLS sales within a 0.5-mile radius of the subject property. Properties must share the same broad type (single-family, condo/TIC, multi-family) and must have sold within the past six months. Outliers — sales more than two standard deviations from the median price-per-sqft — are removed before computing the average, so a single distressed or inflated sale does not skew the result. Recent sales are weighted more heavily than older ones using exponential decay (a sale from eight months ago carries roughly half the weight of a sale from last week), so the fair value tracks the current market rather than trailing conditions.

HomeBidder applies a log-linear hedonic model grounded in Bay Area empirical research. For lot size, an elasticity of 0.15 is used — meaning doubling the lot size relative to a comp adds roughly 10.4% to value (ln(2) × 0.15). For interior square footage, an elasticity of 0.25 is applied. Using logarithms rather than simple linear scaling prevents runaway adjustments on extreme size differences and better reflects the diminishing returns of extra space in dense urban markets. Lot adjustments are capped at −15% / +20% and sqft adjustments at −20% / +20%.

The fair value is always accompanied by a low–high confidence interval. The interval widens when: there are few comps (< 3), the comp pool shows high price dispersion (coefficient of variation > 20%), the property is missing square footage or lot size data, a fallback valuation method (price-per-sqft or list price) was used, or the property has no list price at all (+3% for unlisted properties). The interval narrows as more matching comps are available and their adjusted prices converge. Missing size data also immediately triggers a 'low confidence' flag on the analysis.

Yes. When no list price is available — for instance, an off-market property or a pocket listing shared directly by an agent — HomeBidder derives fair value entirely from comparable sales or price-per-square-foot data. The offer range, risk flags, and investment analysis all work the same way. The confidence interval is widened by 3% to reflect the extra uncertainty of having no list price anchor, and the analysis is labeled as 'Not listed' rather than showing a blank price.

The mid offer is the fair value estimate itself. The low offer subtracts a margin that reflects local market conditions and the property's risk profile — conservative buyers or high-risk properties get a larger cushion. The high offer adds a premium calibrated to recent overbid rates in the same neighborhood; in hot SF submarkets this can be 5–15% above fair value. The spread between low and high is not fixed — it grows with uncertainty (wider confidence interval) and shrinks when comps are dense and consistent.

Tenancy-in-Common (TIC) properties in San Francisco typically sell at a 7% discount to equivalent condo or single-family values. This discount reflects two structural differences: (1) TIC financing is harder to obtain — most lenders require fractional loans at higher rates — and (2) TIC ownership carries liquidity risk because resale depends on co-owner cooperation or partition. HomeBidder detects TIC ownership signals in the listing description and property type, and applies the −7% discount to the comp-derived fair value before computing the offer range.

For whole-building multi-family properties (duplexes, triplexes, small apartment buildings), HomeBidder adds an income-based premium on top of the comp-derived fair value. The premium uses a Gross Rent Multiplier (GRM) of 18 — a commonly used benchmark in the Bay Area — applied to the estimated annual rent for all units. For example, a duplex with two units each renting at $3,500/month has annual gross rent of $84,000; at GRM 18 the income value is $1,512,000. The difference between that income-capitalized value and the comp-derived value is added as a premium, capped at 10% of fair value, so a single outlier rent estimate cannot dramatically distort the offer range.

HomeBidder evaluates the following risk categories: FEMA flood zone designation, CAL FIRE fire hazard severity zone, seismic liquefaction zone, earthquake fault proximity (Alquist-Priolo zones), CalEnviroScreen air quality (PM2.5) and environmental contamination percentiles, highway noise proximity, TIC ownership structure, multi-family structure type, and tenant-occupied status detected from the listing description. Each factor is rated high, moderate, or low and displayed in the Risk tab.

HomeBidder queries the FEMA National Flood Hazard Layer for the property's coordinates. Properties in Zone AE, AO, or VE are flagged as high flood risk (Special Flood Hazard Areas where flood insurance is typically required for federally backed mortgages). Zones X-shaded (moderate) and X-unshaded (minimal) are rated accordingly.

Fire risk is sourced from CAL FIRE's Fire Hazard Severity Zone (FHSZ) data. Zones are classified as Moderate, High, or Very High in State Responsibility Areas, and Local Responsibility Area classifications are also incorporated. Properties in Very High FHSZ zones receive a high risk rating; High zones receive moderate; Moderate zones receive low.

Projections use FHFA House Price Index data specific to the property's metropolitan statistical area (MSA) and zip code where available. The historical compound annual growth rate (CAGR) over the most recent full decade is used as the base appreciation rate. The model compounds this rate forward from the estimated purchase price. These are illustrative projections — actual appreciation depends on market conditions, interest rates, and local policy changes.

Registered users receive a property-specific rent estimate from RentCast's automated valuation model, which is neighborhood-aware and accounts for bedrooms, bathrooms, square footage, and property type. Anonymous users receive a Census ACS zip-code median rent for the property's zip. The source of the rent estimate is always labeled in the Investment tab.

The monthly ownership cost includes four components: principal and interest (at the current 30-year fixed rate), property tax at 1.2% of purchase price (Prop 13 base rate plus typical Bay Area supplemental assessments), homeowner's insurance at 0.35% of purchase price, and ongoing maintenance at 0.5% of purchase price annually. This all-in cost is compared to the estimated market rent for a similar unit. The comparison is displayed at the 10, 20, and 30-year horizon to show how equity accumulation changes the long-run cost picture.

HomeBidder uses CAASPP (California Assessment of Student Performance and Progress) proficiency rates — the percentage of students meeting or exceeding state standards in Math and English Language Arts. The nearest elementary, middle, and high school within 2 miles are shown, color-coded green (≥ 60% proficient), yellow (40–59%), or red (< 40%). Data is from the 2022–23 CAASPP administration and covers 31 Bay Area schools in the built-in dataset.

HomeBidder finds the nearest BART station, Caltrain station, and MUNI Metro stop to the property and reports the straight-line distance to each. A transit premium signal fires when the nearest rapid-transit stop (BART or Caltrain) is within 0.5 miles. Walking distance to transit is a well-documented value driver in the Bay Area; properties near BART in particular have historically commanded a measurable premium.

For San Francisco properties, HomeBidder queries the DataSF Socrata API (SFPD incident data) for incidents within a 0.5-mile radius over the past 90 days. For other Bay Area cities, the SpotCrime API is used when a key is configured. Incidents are split into violent crime (assault, robbery, homicide, rape) and property crime (theft, burglary, auto theft, arson). The counts are color-coded and displayed in the Risk tab alongside the top crime types reported.

When a property is detected as a fixer (based on listing description keywords and MLS condition signals), HomeBidder uses an LLM to generate a line-item renovation estimate tailored to the property's age, size, and description. Line items are drawn from a curated list of renovation categories typical for Bay Area homes — roof replacement, electrical panel upgrade, plumbing, kitchen and bath remodels, foundation work, seismic retrofitting, HVAC, windows, and siding. Each line item can be toggled on or off to adjust the total estimate to your scope.

Renovation estimates are ballpark figures intended to calibrate your offer, not contractor bids. Bay Area construction costs vary significantly by neighborhood, contractor availability, permit complexity, and material choices. The estimates are based on median contractor rates for the SF Bay Area and are most reliable for full-scope renovation of a typical 1900–1970s SFH. Always obtain at least two licensed contractor quotes before finalizing your offer on a fixer property.

Yes. After an analysis is complete, you can upload a PDF inspection report from the analysis detail page. HomeBidder sends the report to Claude as a document, which parses it into structured findings organized by category (roof, electrical, plumbing, HVAC, foundation, etc.) along with severity ratings and recommended actions. These findings then replace the keyword-based fixer estimate with a scope derived directly from the inspector's observations — typically more accurate for properties where you have already conducted due diligence. The individual line items remain toggleable so you can include or exclude work from the total.

Every analysis you run is automatically saved to your history. Registered users can view all past analyses from the History page; anonymous users see analyses saved in their current browser session. You can star any analysis with the heart icon to mark it as a favorite — starred analyses are sorted to the top of your history and persist across sessions for logged-in users. The history page also shows the address, date, fair value estimate, and whether the property was marked as seen.

Mark Seen lets you record a first-hand impression of a property after visiting it in person. Available to logged-in users from any analysis detail page, it captures two ratings: a quality rating (terrible / bad / neutral / good / excellent) reflecting the property's overall physical condition, and a location rating (bad / neutral / good) reflecting the neighborhood feel. HomeBidder combines these into a composite score — (quality + location) / 2 — displayed alongside the analysis in your history. This gives you a structured way to compare multiple properties you've toured, beyond what the automated analysis can observe.