Comps — comparable sales and comparable rents — are how the market tells you what a property is worth and what a unit should rent for. The data is easy to find; the skill is reading it honestly. Cherry-picked comps are how buyers overpay and landlords overprice, one flattering data point at a time.
What makes a comp comparable
Closer on these dimensions beats more data points: location (same neighborhood or school zone — a half-mile can be a different market), property type and size (bedrooms, baths, square footage within ~20%), condition and vintage (renovated vs original is often a 10–20% spread), and recency (sales within 6 months, rents within 3 — older is stale in a moving market). Five genuinely similar data points beat twenty loose ones.
Adjust, don't average
Comps are never identical, so adjust each toward your subject: subtract for the comp's extra bath or garage, add for your renovated kitchen or in-unit laundry. Keep adjustments few and honest — if a comp needs five adjustments, it isn't a comp. For rents, weight units actually RENTED over asking prices; asking rents that sit for weeks are evidence of the ceiling, not the market.
Where to get them
Sales: recently closed (not listed) prices from listing portals, county records, or an agent's MLS pull — agents will often run comps hoping for future business. Rents: listing sites for asking data, a local property manager for what's actually signing, and your own vacancy experience as the final check. Note the source's bias: a listing agent's comps justify the price; an appraiser's comps must survive review.
Worked example
Valuing a 3BR/2BA duplex side: comps at $255,000 (similar, but one fewer bath: adjust +$8,000), $271,000 (renovated, yours isn't: −$15,000), and $262,000 (nearly identical, closed 6 weeks ago: no adjustment). Adjusted: $263,000, $256,000, $262,000 — a tight $256,000–$263,000 band. The $290,000 "comp" from the luxury block eight blocks over gets excluded, not averaged in. That exclusion is the discipline.
When this rule of thumb breaks
Thin markets break comps: rural areas and unusual properties may have three loosely relevant sales a year — widen the radius and time window, but treat the answer as a range with error bars, and lean harder on the income approach. Fast markets break them differently: six-month-old comps in a market moving 1%/month are systematically 6% stale. Distressed sales, family transfers, and off-market deals contaminate the record. And comps describe the market as it was — a new employer, a zoning change, or a wave of construction changes what comes next, which no comp can show.