§ 02 / METHOD
twelve comps.
six adjustments.
one number.
PlainComp is a transparent, data-driven property valuation tool. No black box, no proprietary mystery — every step of the model is documented below.
01 / OVERVIEW
PlainComp is a free, instant property valuation tool for residential properties across the United States. It produces comparative market analysis estimates by combining recent comparable sales data with a statistical model and AI-generated narrative analysis.
Designed for homeowners researching value, buyers evaluating purchases, investors screening opportunities, and anyone who wants a quick, defensible estimate without waiting for a formal appraisal.
Operated by Matt Rybicki, a licensed Arizona REALTOR (ADRE SA712773000) with Keller Williams Arizona Realty. Not affiliated with any lending institution or appraisal firm. Valuations are automated estimates for informational purposes only.
02 / MODEL
When you enter an address, PlainComp runs through a multi-step pipeline. Every step is deterministic and inspectable.
Comparable Sales Retrieval
The model queries the Rentcast API to pull recent property sales within a defined radius of the subject. Returns raw comparable sales including prices, dates, square footage, beds/baths, lot sizes, and coordinates.
Sanity Checks and Filtering
Raw comps are filtered through quality gates. Sales missing critical data are removed. Square-footage gating drops comps drastically different in size from the subject. PPSF outlier detection removes anomalous pricing that would skew the estimate.
Ridge Regression Adjustment
A ridge regression is fit on the comparable set to estimate the relationship between square footage and sale price. Each comp is then adjusted to what it would theoretically sell for at the subject's size. Ridge is used over OLS because it handles small samples and multicollinearity gracefully via regularization.
Composite Scoring
Each comp gets a composite quality score: 30% recency, 40% size similarity, 30% bed/bath match. Recent sales reflect current conditions; size dominates because it is the strongest price predictor; layout match captures functional similarity.
Distance Weighting
An exponential decay function is applied based on geographic distance from the subject. Closer comps contribute more. Reflects the reality that hyper-local factors — street, subdivision, immediate neighborhood — meaningfully impact value.
Top Comp Selection and Outlier Removal
Comps are ranked by combined composite × distance weight. Top 10 are selected. An iterative outlier pass then removes any remaining adjusted-price outliers within the final set, ensuring the estimate is based on a tight, coherent group.
Weighted Valuation
The final estimated value is a weighted average of the adjusted prices of the selected comps, where weights combine the composite quality score and distance weight. The most similar, most recent, closest comps have the greatest influence.
03 / GRADES
Every valuation reports two grades that quantify how much weight to give the number.
A — CONFIDENCE
How tight is the estimate
- COMP COUNT
More comps = stronger statistical basis. 8+ is more reliable than 3. - PRICE SPREAD
Tight cluster of adjusted prices = strong consensus. Wide spread = more uncertainty. - PROXIMITY
Geographically close comps provide more relevant pricing signal than distant ones.
B — ROBUSTNESS
Quality of the underlying data
- CV — DISPERSION
Coefficient of variation. Low CV = consistent pricing. High CV = heterogeneous. - HHI — CONCENTRATION
Herfindahl-Hirschman Index. Low HHI = balanced weights. High HHI = single-comp dominated. - ADJUSTMENT BURDEN
How much the regression had to push prices around. Less adjustment = naturally comparable sales.
04 / NARRATIVE
Each report includes a narrative summary generated by Anthropic's Claude. The model receives the valuation data, comp details, confidence metrics, and robustness metrics, and produces a plain-language interpretation.
The narrative highlights key patterns, explains the basis for the estimate, surfaces concerns, and contextualizes the grades. The AI has access only to what the model provides — read it as an interpretation of the data, not independent analysis.
05 / VS ZESTIMATE
Zillow and Redfin use massive proprietary datasets, MLS feeds, and ML models trained on millions of transactions nationwide. They are powerful tools with broad coverage. PlainComp takes a different approach.
01
TRANSPARENT METHOD
The entire model logic is on this page. You know exactly how the number was derived. Zillow and Redfin treat their models as proprietary black boxes.
02
COMP-LEVEL DETAIL
Every comparable sale that contributed, its weight, its adjusted price, its distance. Evaluate the comps yourself.
03
QUALITY METRICS
Confidence and robustness grades give you a structured way to evaluate trust, not just a single number.
04
AI INTERPRETATION
A narrative analysis explains the estimate in context — not just numbers and charts.
05
NO ACCOUNT
No sign-up, no data collection, no lead generation. Enter an address, get a result.
06
TRADEOFF
Narrower dataset (Rentcast vs. MLS) and a simpler model. For a quick, transparent estimate it fills a useful niche.
06 / LIMITATIONS
No automated valuation is a substitute for professional judgment. PlainComp has specific limits.
DATA COVERAGE
Works best in markets with high transaction volume. Rural areas with few sales produce less reliable estimates.
NO INTERIOR ASSESSMENT
Cannot evaluate renovations, upgrades, damage, deferred maintenance, or anything requiring physical inspection.
DATA DEPENDENCY
Only as good as the comp data available from Rentcast. Sparse markets may produce weak estimates or none at all.
RESIDENTIAL FOCUS
Designed for SFH, condos, townhomes, small multi-family. Not for commercial, industrial, agricultural, or vacant land.
LAGGING INDICATOR
Relies on closed sales — reflects where the market has been, not where it is going. Lags in rapidly shifting markets.
NO FORECASTING
Does not predict future values, appreciation rates, or market trends.
UNIQUE PROPERTIES
Custom builds, historic homes, properties with unusual features may not have meaningful comparables.
For important financial decisions, always consult licensed professionals. See the valuation disclaimer for full details.
07 / FOUNDING 20
The first ten paid Pro signups and first ten paid Investor signups become founding members. Their price is locked forever, and their seat in the cohort is permanent.
Founding membership is open through public launch. First 10 paid Pro signups and first 10 paid Investor signups become founders.
▮ founding cohort
Founding membership opens with public launch. By invitation through the early phase. The valuation engine is already trained on 730,000 real Maricopa County public-records sales going back five years — that's the credibility you sign up for, not a claimed-seat counter.
Want a seat? View Pro plans →