More than half of GTA relists go back on the market the day the old listing ends

What we measured

When a home does not sell, the listing is often withdrawn and immediately entered again under a new MLS® number. To the board’s statistics the second entry is a new listing, and its days-on-market counter starts at zero. Nothing about the house has changed.

TRREB publishes new listings every month. It does not publish terminations, so the share of “new” listings that are the same homes coming back is not a number anyone can look up. We reconstructed it from listing-level status history: every listing in the feed matched to a property key, and every property’s full sequence of listings rebuilt in order.

The board did publish the tell itself. In its July 2026 release, days on market measured per listing rose 6.7% year over year while the same figure measured per property rose 12.5% — the gap between a listing and a home widening, reported at region level and left unexplained. This is that gap, decomposed to municipality and property type.

A home is on the market twice as long as it reports

Across 122,975 GTA sales, the median reported days on market is 20. Measured from the first time the property came to market — counting the listings that were withdrawn along the way — the median is 40.

The average conceals two different populations. On the 62% of sales that happened on the first listing, the two numbers are the same. On the 38% that took more than one listing, the reported median is 22 days and the true median is 107 — a factor of 4.9. The longest chain in the corpus is 35 consecutive listings of the same home.

The timing is the finding

A seller who genuinely reconsiders does not come back the same afternoon. Of 146,210 GTA relists, 58.3% were entered on or before the day the previous listing ended, 67.7% within a day, and 77.2% within a week. We make no claim about intent — the distribution speaks for itself.

What relisting buys

Relisting is mostly a price decision. 75.5% of relists came back at a lower asking price, at a median cut of 4.79%; 10.7% came back higher. The rest returned at the same price, which buys nothing but a fresh clock.

How much of the market this is

In July 2026, 44.2% of GTA for-sale new listings were properties coming back rather than homes arriving for the first time — 6,586 of 14,900 in this corpus.

Read this figure with its denominator. It divides by our count of new listings, which runs 2.9% above the board’s published figure for July 2026 and 2.6% above it a year earlier — close, stable, and not the board’s number. Our sales counts, average prices and per-listing days on market all reproduce the published figures within a few percent; the Method section sets out all of it line by line. The days-on-market and timing findings above depend on no denominator at all and are unaffected either way.

The trend rose through 2025, peaked in January 2026 and has eased since. One caution about that peak: listing agreements commonly expire at year end, so December and January carry a genuine seasonal bulge in expiries and therefore in January relists. With two years of data there is only one January, so the chart below plots each year separately against a shared calendar rather than as a single line.

Where it happens most

The practice is not evenly spread. Among the larger markets, Brampton sits near the top at 47.6%, while the City of Toronto is near the bottom at 41.9% — the opposite of what most people assume. The smaller municipalities at the very top of the table have correspondingly small monthly counts, so read them as indicative rather than precise.

Method

This section is longer than a summary needs to be, on purpose. The one previous public attempt at this number matched addresses in a spreadsheet for a single month, and the difference between that and this is entirely in the method.

Data

1,208,682 listing records from TRREB’s VOW feed covering 2024-08-16 to 2026-08-16, the board’s full retention window. Each record carries its own status and status dates, which is what makes the reconstruction possible; the published monthly statistics are aggregates and cannot be decomposed this way.

Matching listings to properties

Listings are grouped by a property key built from unit number, street number, direction, street name, suffix and postal code — so a condo unit is distinguished from the unit above it, which is the failure mode that makes address matching unreliable in apartment stock. 40.2% of records carry a unit number and 99.99% of those resolve into the key. The postal code is present in every key, and the last-resort fallback that would key a listing by its own MLS® number — joining it to nothing — fires on 0.0% of records.

The match is independently verified. TRREB prefixes every MLS® number with a letter encoding the board district, and nothing in our key derives from it, so two listings we claim are the same property must still agree on that letter. Across the 344,984 matches we claim anywhere in the corpus they agree 99.96% of the time (135 disagreements); across the 146,210 that the findings on this page are drawn from, 99.97%. The disagreements are dominated by Toronto Central/West pairs — a boundary effect, not a random collision.

What counts as a relist

A listing entered within 180 days of a previous listing of the same property, of the same transaction type, ending in a withdrawn state. Halving that window to 90 days moves July 2026 from 44.2% to 42.8% and leaves the shape of the series unchanged — the rise through 2025, the January 2026 peak and the easing since are all still there — so nothing on this page turns on that choice.

Withdrawn means Terminated, Expired or Suspended, treated as one class. That is a rule rather than a preference. Of the 60,916 GTA listings that ended in July 2025, 0.0% were coded “Suspended” and 36.9% “Terminated”; a year later, of 58,821, Suspended was 3.5% and Terminated 32.2%. The status is new and it is taking volume from termination, so a measure counting terminations alone would have reported a 2026 decline that is a change in coding, not in behaviour.

Where two listings of the same property genuinely overlap by more than three days they are treated as coexisting, not sequential, and excluded — most such pairs are one property offered for sale and for lease at the same time.

Two classes are excluded throughout: vacant land, and the feed’s non-residential catch-all. The second is the one that matters. It holds commercial, industrial, office and retail listings, businesses for sale, parking spaces and part-floor rentals, and it is not what TRREB counts as a residential new listing either. It is also not stable: it ran 201 GTA records in July 2025, 1,216 in July 2026 — 1.1% of for-sale records rising to 7.5% — as coverage of that part of the market widened. Those records relist on a different rhythm and would distort both the level and the trend. Put both classes back in and July 2026 reads 41.8% instead of 44.2%, with the timing finding unchanged at 58.1%.

Days on market

Reported days on market is the figure carried by the listing a home sold on. True days on market counts from the first listing of the same property through to the sale. Sale dates use the date the sale was entered, not the closing date — closing is a legal event weeks later, and counting on it puts July 2026 GTA sales at 7,234 against the board’s 5,995 (+20.7%) and turns a flat year into a rising one.

We would rather have used the board’s own field than computed one. TRREB’s VOW feed declares a cumulative days-on-market field in its schema, so we collect it — and it arrives empty on every record we have received, all 1.2 million of them. A field added to our pipeline in the same change, the date a listing was suspended, populates on tens of thousands of rows, so this is the feed rather than our reading of it. There is nothing to compare our figure against, and that is part of why the property-level view of this market is not already published.

How this corpus compares with the board’s published figures

Reconciled against TRREB’s published Market Watch over the board’s own jurisdiction, on exactly the population every figure on this page is drawn from — Greater Toronto, for sale, excluding the two classes above — the corpus tracks the board closely, and its residual is stable across the two years rather than growing.

The same four lines for July 2025, ours against the board’s: new listings 18,076 against 17,613 (+2.63%), sales 6,151 against 6,100 (+0.84%), average price $1,049,682 against $1,051,719 (-0.19%), average days on market 30.3 against 30.

Two things follow. Our count of new listings runs about 3% above the board’s in both Julys, so every share on this page is a share of a slightly larger denominator, and we describe it that way throughout. And the year-over-year change it implies — -17.6% against the board’s published -17.8% — matches to two tenths of a point, which is the test that matters for a study about how listings are counted.

An earlier cut of this study reported that residual as growing, from 3.77% to 11.27%, and could not explain it. The explanation is the comparison rather than the data: those figures counted the corpus’s 16,116 records of every kind against a published number that counts residential ones. The arriving commercial records described above are the entire difference, and they are excluded from every figure on this page. We are stating the correction rather than quietly dropping it, because the earlier figure has been sent to people.

One thing we still do not do is restate the board’s sales-to-new-listings ratio. Netting relists out of the denominator moves it a long way, and anyone can run that arithmetic from the shares published here, but the result would be our ratio rather than the board’s and we would rather it were labelled that way.

Against the board’s own days-on-market fields

TRREB publishes two: LDOM, which measures one listing, and PDOM, which measures the property. Our per-listing average reproduces the board’s LDOM to within 1.1 of a day in July 2026 (33.1 against 32) and 0.3 of a day a year earlier (30.3 against 30), which is the strongest single check on the days-on-market work here.

Our property-level figure runs above the board’s PDOM — an average of 87.7 days in July 2026 against the published 45 — and the two are not computed the same way, so the difference is not a discrepancy to reconcile. We chain a property through every relist that falls within 180 days of the one before it, however many that is and whoever listed it; the board’s field is maintained within its own listing history. Read the published PDOM as a floor and this as the fuller count. It is also why the board’s July release is the tell this study starts from: its PDOM rose 12.5% year over year while its LDOM rose 6.7%.

Scope

Figures are the Greater Toronto Area, matching TRREB’s jurisdiction. Relisting practice is shaped partly by how individual boards handle listing agreements, so comparisons that cross board boundaries measure some of that alongside market behaviour, and we have kept this study inside one board. For contrast within it, the lease market behaves very differently — 14.8% of July 2026 new lease listings were relists against 44.2% for sale — which is also a check that the measure is not simply picking up general churn.

Use this

The charts and figures on this page are free to reproduce with a link back to this page. The municipality-level data behind them is downloadable as a CSV — one row per municipality, carrying both the July 2026 relist share plotted above and the full-window medians behind the days-on-market chart, so every figure on this page can be recomputed from it.

Questions, or a cut of the data specific to your area? Get in touch — we are happy to run one.