Two widely syndicated sources reported the number of Rochester NY homes sold in July 2026. One said 509. One said 643. That is a 26 percent gap in a single, supposedly countable fact, and it is not a rounding problem.

This post walks through exactly where those two numbers came from, why neither survives basic arithmetic, and what else on those same pages falls apart when you check it. By the end you will have a short audit routine you can run on any monthly market report in about five minutes. You will also know which Rochester figures I do trust right now, and there are fewer of them than you would hope.

The two counts, side by side

Start with the second row, because it fails on its own terms. 643 is not up from 707. It is down 9.1 percent. The sentence contradicts its own numbers. Whatever process produced that page did not have a human checking the direction of the comparison.

The first row fails differently. A 78.6 percent year over year jump in monthly closings implies July 2025 produced only about 285 sales in the same geography. Monthly closings in a market this size do not organically move that far. Closings in July reflect contracts written in May and June, and there was no May or June event that would double the pipeline. A move that large almost always means the data set changed: more ZIP codes pulled in, a different property type filter, a feed that was partially broken in the prior year, or a vendor switching providers. It is a coverage change wearing the costume of a demand surge.

So one figure is arithmetically self-contradicting and the other is implausible on its face. My position is that neither belongs in a decision. When two sources disagree this badly and both show internal defects, the correct output is "unknown," not "pick the friendlier one."

Half the problem is that "Rochester" means three different places

Before you can count sales, you have to agree on the boundary. These sources do not. Look at what happens to the median price for roughly the same moment.

That is a spread of $199,000 to $319,900, a 60.8 percent gap, for a market that is being described with the same word. Some of these pages mean the city. Some mean the county. Zillow's figures are a modeled index, not an average of actual closings, so do not line them up against sale prices as if they measure the same thing.

Once you accept that the denominator is unstable, the sold-count disagreement stops being mysterious. 509 and 643 may both be internally consistent counts of two different footprints, neither of which is labeled clearly enough for you to know which one you are holding.

The days on market tell

Here is the check I run first, because it is fast and it catches most bad pages. Pull every pace and supply figure on the report and see whether they can coexist.

• Zillow, Monroe County, as of April 30, 2026: homes go pending in roughly 8 days.
• Movoto, Rochester, July 2026: 13 days on market, against 13 days a year earlier.
• Redfin, a northwest Rochester neighborhood, three months ending May 2026: 11 days, against 11 days a year earlier.
• Houzeo, Rochester, July 2026: 51 days.

Three sources cluster at 8 to 13 days. One says 51. And the 51 contradicts its own page twice over. Houzeo's own months of supply figure for July 2026 is 0.33, which converts to about 10 days of supply. Houzeo also publishes its own interpretation rule: under 45 days signals a seller's market, 45 to 70 days indicates a balanced market, over 70 days favors buyers. Its own 51 day print therefore classifies Rochester as balanced while the rest of the page calls it a strong seller's market.

That is not a debatable judgment call. It is a page arguing with itself. Treat 8 to 13 days as the credible range and treat 51 as an artifact.

Two numbers that look like corroboration and are not

The same July 2026 report shows 169 homes for sale and 0.33 months of supply. Those look like two independent facts pointing the same way. They are not. 509 times 0.33 equals 168, and 169 divided by 509 equals 0.332 months. One figure was derived from the other. Citing both as mutual confirmation doubles the apparent evidence without adding any.

Meanwhile Resideline reports 56 active listings, 68 pending, for July 2026. So we now have 169 and 56 as active listing counts for the same place and month. Both are implausibly low for a city of roughly 210,000 and are almost certainly filtered subsets, likely single family only or a partial ZIP set. Note them, do not repeat them as gospel.

One more single-source figure worth naming: a median sale to list ratio of 120.57 percent, July 2026, Houzeo. Rochester genuinely runs high over-list premiums by national standards. A median of 120.57 percent is still extraordinary and no second source corroborates it. If you use it, attribute it and caveat it.

What I actually trust this month

Mortgage rates. They come from one publisher on a fixed weekly schedule and they are not scraped from anyone's listing feed.

• 30 year fixed: 6.67 percent, week of August 13, 2026, Freddie Mac PMMS. Prior week: 6.69 percent. A year earlier: 6.58 percent.
• 15 year fixed: 5.96 percent, week of August 13, 2026. A year earlier: 5.71 percent.
• 10 year Treasury: 4.668 percent on August 13, 2026, up from 4.656 percent.

That is 9 basis points higher year over year on the 30 year. On a $300,000 loan that is roughly $18 a month. Whatever moved Rochester prices over the past year, financing cost is not the story. That drop from 6.69 to 6.67 also ended a run of five consecutive weekly increases.

Reporting lag: there are no August closings yet

If you see a piece published in mid August claiming to report August closings for Monroe County, close the tab. August closings are not published until early to mid September. As of mid August 2026 the freshest local sales data is July 2026 from aggregators, and the freshest primary-source price is May 2026 from NYSAR by way of the Rochester Business Journal. The only genuinely August figure available is the mortgage rate.

The five minute audit

• Find the boundary. City, county, metro, or an unnamed ZIP set. If the page will not say, the number is not usable.
• Check the direction of every year over year claim. "643, up from 707" is the kind of error that survives publication because nobody subtracts.
• Convert months of supply into days. Then compare it to the stated days on market. If they disagree by a factor of five, one of them is wrong.
• Look for derived figures posing as independent ones. Multiply the sales count by months of supply and see if you get the inventory count.
• Ask whether a big year over year swing has a mechanism. No mechanism, no story. Methodology change is the default explanation.
• Separate modeled indexes from closed sales. A home value index and a median sale price are different measurements.

Why the national headlines feel wrong here

A mid-year 2026 review described the market as unwinding from the pandemic-era frenzy, with buyers having more choice than at the start of the year and modest price growth. The same piece notes that parts of the Midwest, Northeast and Bay Area have become considerably more expensive as buyers compete over dwindling inventory, while many Sun Belt cities cooled and listings sat for months. The cooling narrative is largely a Sun Belt narrative. When a Redfin-derived page says Rochester's market "has remained very slow," that is a national model output printed on a local URL. Redfin's own competitiveness score for northwest Rochester was 92 out of 100, most competitive.

Also worth labeling clearly: Houzeo projects Rochester prices to rise 2 to 4 percent in 2026. That is a vendor forecast, and it sits oddly next to the same vendor's 17.95 percent year over year print for July 2026.

What this means if you are buying or selling in south and southeast Monroe County

None of the figures above describes Henrietta, West Henrietta, Pittsford, Brighton, Mendon, Rush or Honeoye Falls. They describe the city of Rochester or the county as a whole. Publishing a city-wide months of supply figure as though it explains a Pittsford closing would be the same mistake this post is about. The relevant number for a specific house is the recent comparable sales on its street, in its price band, in its condition. That takes a search built for the property, not a dashboard.

If you want the actual comparables behind a purchase or a listing decision, and a plain explanation of where each figure came from, book a time with me. I will show you the pull, the filters and the dates, and I will tell you when the data does not support an answer.

Khem Kadariya

About this data

The figures in this post were compiled from publicly available sources including Houzeo, Movoto, Redfin, Zillow and FRED (Federal Reserve Bank of St. Louis), along with other public market data. Real estate numbers change quickly, and these were accurate as of August 2026. For current figures on a specific home, street, or town, ask me directly rather than relying on a published average.