A Halifax seller cited a rising national average price to argue for a higher list price, while CREA’s benchmark index for the same period was falling — both figures were accurate, and the agent had to explain why they disagreed.
At a glance
A Halifax seller preparing to list had read a national news story built around one line from a monthly housing report: the average home price was up year over year. He wanted to price above his agent’s recommendation on the strength of that headline. His agent’s own read of the market, based on the same underlying CREA release, was that prices were actually still cooling — the two positions looked contradictory from the outside, but both were reading real numbers correctly.
CREA’s own national statistics package explains exactly why an average price and a benchmark price can move in opposite directions from the identical set of transactions: “Average or median prices can change a lot from one month to the next and paint an inaccurate or even unhelpful picture of price values and trends.” An average is simply total sale value divided by the number of sales — if a run of larger, higher-end properties happens to close in a given month, the average rises even if typical home values across the market are flat or falling.
This was not a hypothetical tension for this seller’s file — it was visible in the exact same CREA release he had read the headline from. The release he cited reported the non-seasonally adjusted national average home price at $674,819 in July 2026, up 0.2% from the same month a year earlier, while the MLS® Home Price Index — in the very same release — “edged up 0.1% month-over-month and was down 3.3% on a year-over-year basis.” Two real, correctly-reported figures, from the same month, pointing in opposite directions.
National average price, July 2026: $674,819, up 0.2% year over year.
MLS® HPI Composite, same release: down 3.3% year over year, up 0.1% month over month.
National sales-to-new-listings ratio for the same month: 51.3%, inside CREA’s own stated balanced-market band of roughly 45% to 65% — a third figure the seller had not seen, and one that supported the agent’s cooling read rather than the average-price headline.
CREA built the HPI specifically because the average is structurally unreliable for exactly this kind of month-to-month comparison. The methodology page states the HPI uses “over 15 years of sales data and statistical modelling” to define a “typical” benchmark home based on the features of properties actually bought and sold, monitored by neighbourhood and housing type every month — giving “apples to apples” comparisons that a raw average cannot, because the average has no way to normalize for what mix of properties happened to sell.
For this seller, the practical translation was straightforward: the national average had been lifted by a run of higher-end sales elsewhere in the country, not by a genuine market-wide price increase. The HPI, tracking a consistent property type over time, showed the underlying trend the average was obscuring — and it was the figure that actually matched what recent comparable sales in the seller’s own neighbourhood were doing.
The seller agreed to price against the HPI trend and the local comparable sales his agent had pulled, rather than the national average headline. The home listed within the agent’s original recommended range and sold in nineteen days. The agent now keeps both figures — the current average and the current HPI reading, sourced from the same CREA release — in every listing presentation specifically so a client arriving with one headline number has the other one in front of them before the pricing conversation starts.
The tell was that the seller’s headline and the agent’s market read were both, individually, defensible — which is exactly what made the disagreement hard to resolve by assertion alone. Neither side was wrong about the number they were quoting. The only way through was showing the seller both figures side by side, from the identical release, so the gap itself became the explanation rather than a dispute over whose source was more credible. A client is far more likely to accept a pricing strategy once they can see why two real numbers disagree, rather than being told to simply trust the agent’s figure over the one they found themselves.
Related reading: the glossary entry laying out benchmark price versus average price in full, and a related file where the wrong comparables, not a headline, produced the mismatch: a CMA built on the wrong comparables.
A 30-minute call is enough to tell you whether AI pays for itself here.
