How homes are scored
Each home's score (0–100) is a weighted blend of the four pillars below. The sliders set how much each matters — they auto-normalize to 100% and the list re-ranks instantly as you drag them.
Waterfront — GPS distance to the real shoreline + water type. Open water & big coves score high; fingers, near-water and inland are excluded.
Value / deal — asking price vs. the consensus of three independent AVMs (Quantarium, Cotality, Collateral Analytics), which a backtest found ~5–10× more accurate than comps alone — even on waterfront. A deal only counts when the AVMs and our list-independent sold-comp model agree it's underpriced; when they disagree the gap is shrunk toward 0, so noisy false bargains drop out. Our comp model is the fallback when a home has no AVM. The gap maps to the score on a smooth absolute curve: priced at the estimate = 50, and a home 0.5% over vs. 0.5% under fair value scores nearly the same — only meaningful gaps move the pillar (±15% → roughly 88/12). Long market time adds a small negotiating-leverage bonus (15% of the pillar).
House & livability — five weighted components: condition 35% (the largest), size 26%, lot 15%, beds 12%, baths 12%. Condition starts from an era curve of the build year (a 2020 build starts ~90, a mid-70s build ~22), then listing cues adjust it: a stated renovation lifts it to roughly the remodel year's era, quality finishes add points, and an old home with no updates mentioned takes an extra penalty — so an unrenovated 1970s house drags this pillar hard, and a true remodel earns most of it back. A stated fixer/as-is caps condition at 25. AI photo reads adjust condition when photos contradict the text.
Dock — counts boathouse, boat/jet-ski lifts, slips, bulkhead and other water improvements named in the listing.
Open “Why this score” on any card to see the exact arithmetic: each pillar's 0–100 score × your current slider weight = its points, plus any flood/gut-call penalties, totalling to the badge score — with the Value and House pillars decomposed line by line.
💡 Suggested offer — advisory, not part of the score. Anchored on a condition-adjusted comp value (comps can't see condition, so a fixer's value is shaved down). A 0–100 seller-motivation score — built from cumulative market time across relists, the size/cadence/recency of price cuts, relist churn, and listing-text cues (estate-sale, motivated, priced-to-sell) plus the list-vs-value gap — sets how far below ask the numbers start. Open is where you start the conversation; Target close is the realistic settle point; Walk-away is the max you should pay, anchored on the value estimate. These are heuristic reads, not acceptance probabilities — use the motivation bullets and the comps to judge each case.
