A comparative market analysis (CMA) is an agent's opinion of a property's likely market value, built from comparable sold and active listings — a bespoke, address-specific estimate, distinct from CREA's index-based benchmark price.
RECO's own regulatory material has no dedicated page on how to build a CMA. RECO regulates disclosure and conduct around price opinions, not the methodology itself — comparable selection, adjustment method, how many comparables, how far back to look are professional practice questions, not a regulatory requirement, and no specific comparable count is attributable to RECO or TRESA.
What TRESA does regulate is when an agent may give that opinion at all. Once a brokerage or representative is in multiple representation, the representative cannot advise on price to offer or accept or terms to include for either side. A CMA prepared before multiple representation begins is unaffected, but the advisory step — telling a client what to actually offer or accept based on it — is off-limits once both sides are represented by the same brokerage or representative.
Agents typically use CREA's own national and board-level HPI context as background, then narrow to specific comparables for the CMA itself (see the benchmark-vs-average and MLS HPI entries below). A CMA also carries no CUSPAP backing, unlike a formal appraisal (see broker price opinion, below).
A seller asks an agent to help price a semi-detached listing. The agent pulls five comparable sales from the past 90 days within a kilometre, adjusts for lot size, a finished basement and a garage, and lands on a suggested list price. If that same agent later also represents a buyer bidding on the property, multiple representation now applies: the CMA can still describe the comparables factually, but the agent cannot advise either side what to offer or accept until the parties choose to proceed without that advice or seek independent counsel.
See also: broker price opinion vs an appraisal and benchmark price vs average price.
A 30-minute call is enough to tell you whether AI pays for itself here.
