One number from 0 to 100. When you see a green 82, it's not a vibe - here's exactly how it's built.
When you see a green “82,” it's not a mood. Behind it we run a home through four checks and add up the points. The biggest check by far is the one that matters most to a buyer: how far below its estimated value is this home actually listed?
We compare the asking price to an estimated market value (an AVM - an automated value estimate). If a second independent estimate is available, we average the two, because two estimates that agree are more trustworthy than one. Listed well under value earns the most points.
Then we add three smaller checks: rent yield (would the yearly rent cover a healthy share of the price?), days on market (a home sitting a long time gives you more room to negotiate), and risk (fire/flood lowers the score a little). Roughly, value counts for about half the score, rent yield about a third, and the last two are smaller signals.
The honest part: if a piece of data isn't loaded yet - say we don't have a rent estimate for a home - we don't fake it. We drop that check and re-balance the score over the checks we do have, and the breakdown tells you so. And if a home has no price and no value at all (common for a brand-new foreclosure), we show “awaiting a value estimate” instead of a fake 100.
Every score opens into a plain breakdown that shows which checks counted, how many points each earned, and which were skipped for missing data. Nothing is hidden - and we never use anything about the people in an area. Only the home's price, value, rent, time on market and natural-hazard risk.