95%
Property type
Right on 1319 of 1388 listings: detached, semi, condo, multi-family, land or cottage.
Accuracy across 6 property types
Accuracy report
How accurate are our listing tags?
Every listing we find is tagged with its property type, the investment strategies it fits, its room counts and its deal signals. Here is how often those tags are right, tested on 1388 real listings that people checked by hand.
Updated October 29, 2026
1388 hand-checked listings
Never used for tuning
90%
Investment strategies
We catch 90 of every 100 strategies a listing fits, from BRRRR to House Hacking.
84% of the strategy tags we show are confirmed
92%
Beds, baths & garages
Wrong on just 7% of stated counts. When unsure, we leave it blank instead of guessing.
1846 of 2013 stated counts right
88%
Deal signals
We spot 88 of every 100 signals like "motivated seller", "sold as-is" or "tenant occupied".
80% of the signals we show are confirmed
Live from our database
3,090
listings tracked right now
2,447
new in the last 30 days
2,823
re-checked at the source this week
1,421
cities and towns in 2 countries
100%
placed on the map
3
listing sources, deduplicated
How to read these numbers
Two questions for every tag
A good tag has to do two things: show up when it should, and stay away when it shouldn't. Here is what that looks like for investment strategies, one dot per 1% of tags.
90%
Found
Of all the strategies the listings really support, how many did we tag? Out of 1203, we tagged 1084. Experts call this recall.
84%
Confirmed
Of the strategy tags we showed, how many did a reviewer agree with? 1084 of 1287. Experts call this precision.
The details
Tag by tag
Hover over or tap any bar for the counts behind it. Every chart has the full numbers in a table underneath.
Investment strategies
1388 listings. A listing can fit several strategies.
FoundConfirmed
As-Is275 listings fit it
Short-Term Rental240 listings fit it
House Hacking232 listings fit it
Value-Add203 listings fit it
Cashflow143 listings fit it
BRRRR61 listings fit it
Fix & Flip49 listings fit it
Show as a table
| Strategy | Listings | Found | Confirmed | Missed | Extra tags |
|---|---|---|---|---|---|
| As-Is | 275 | 92% | 90% | 21 | 27 |
| Short-Term Rental | 240 | 92% | 83% | 20 | 44 |
| House Hacking | 232 | 95% | 87% | 11 | 33 |
| Value-Add | 203 | 80% | 78% | 41 | 46 |
| Cashflow | 143 | 90% | 79% | 14 | 34 |
| BRRRR | 61 | 89% | 81% | 7 | 13 |
| Fix & Flip | 49 | 90% | 88% | 5 | 6 |
Reading the bars: a long teal bar with a shorter indigo one, like Value-Add, means we tag that strategy generously. You'll rarely miss one, but give it a second look.
Property type
Share of listings of each type that we typed correctly.
Detached house686 listings
Condo225 listings
Land160 listings
Multi-family140 listings
Semi-detached122 listings
Cottage55 listings
Show as a table
| Type | Listings | Right | Accuracy |
|---|---|---|---|
| Detached house | 686 | 654 | 95% |
| Condo | 225 | 223 | 99% |
| Land | 160 | 154 | 96% |
| Multi-family | 140 | 137 | 98% |
| Semi-detached | 122 | 102 | 84% |
| Cottage | 55 | 49 | 89% |
Beds, baths & garages
Only listings whose text gives the count. When we're not sure, we leave the field blank instead of guessing.
RightLeft blankWrong
Bedrooms890 listings state it
Bathrooms745 listings state it
Garages378 listings state it
Deal signals
The most common of our 59 signals, on 1310 listings. Signals with fewer than 5 examples are in the table only: too few to judge.
FoundConfirmed
Move-in ready577 listings have it
Recent major upgrades349 listings have it
Near transit244 listings have it
Rural or remote161 listings have it
Energy efficient152 listings have it
Sold as-is / no warranty127 listings have it
Suite potential115 listings have it
Waterfront105 listings have it
Municipal services92 listings have it
Rental income stated84 listings have it
Show all 59 signals as a table
| Signal | Listings | Found | Confirmed | Missed | Extra tags |
|---|---|---|---|---|---|
| Move-in ready | 577 | 95% | 81% | 30 | 132 |
| Recent major upgrades | 349 | 93% | 86% | 25 | 52 |
| Near transit | 244 | 97% | 94% | 8 | 14 |
| Rural or remote | 161 | 76% | 83% | 38 | 26 |
| Energy efficient | 152 | 93% | 90% | 11 | 16 |
| Sold as-is / no warranty | 127 | 95% | 90% | 6 | 14 |
| Suite potential | 115 | 84% | 70% | 18 | 41 |
| Waterfront | 105 | 94% | 65% | 6 | 53 |
| Municipal services | 92 | 75% | 75% | 23 | 23 |
| Rental income stated | 84 | 86% | 72% | 12 | 28 |
| Seller financing | 82 | 93% | 91% | 6 | 8 |
| Tenant occupied | 67 | 91% | 74% | 6 | 22 |
| Septic system | 64 | 97% | 85% | 2 | 11 |
| Well water | 64 | 94% | 91% | 4 | 6 |
| Severance / lot split | 61 | 82% | 89% | 11 | 6 |
| Cosmetic work needed | 57 | 60% | 74% | 23 | 12 |
| Vacant on closing | 55 | 98% | 64% | 1 | 30 |
| Motivated seller | 52 | 77% | 65% | 12 | 22 |
| Price reduced | 50 | 90% | 67% | 5 | 22 |
| Major renovation needed | 42 | 62% | 61% | 16 | 17 |
| Cash buyers only | 39 | 100% | 89% | 0 | 5 |
| Leasehold / land lease | 36 | 92% | 94% | 3 | 2 |
| Contract assignment | 30 | 90% | 93% | 3 | 2 |
| Separate meters | 29 | 86% | 78% | 4 | 7 |
| Zoning upside | 25 | 92% | 57% | 2 | 17 |
| Short-term rental allowed | 25 | 92% | 56% | 2 | 18 |
| Healthy reserve fund | 24 | 92% | 85% | 2 | 4 |
| Easement / right of way | 24 | 67% | 73% | 8 | 6 |
| Approved plans or permits | 23 | 83% | 66% | 4 | 10 |
| Garden suite / ADU potential | 20 | 50% | 71% | 10 | 4 |
| Legal suite | 19 | 79% | 94% | 4 | 1 |
| Private, seasonal or water access | 18 | 67% | 75% | 6 | 4 |
| Power of sale / foreclosure | 18 | 61% | 92% | 7 | 1 |
| Wetland or conservation area | 16 | 81% | 81% | 3 | 3 |
| Teardown / lot value | 16 | 44% | 100% | 9 | 0 |
| New-home warranty | 15 | 100% | 100% | 0 | 0 |
| Roof needs work | 15 | 87% | 87% | 2 | 2 |
| No services / off-grid | 13 | 100% | 77% | 0 | 4 |
| Estate sale | 11 | 91% | 100% | 1 | 0 |
| Assumable mortgage | 11 | 82% | 90% | 2 | 1 |
| Agricultural zoning | 11 | 64% | 70% | 4 | 3 |
| Unpermitted work | 10 | 50% | 71% | 5 | 2 |
| Next to rail, highway or power lines | 10 | 40% | 80% | 6 | 1 |
| Oil heating | 9 | 89% | 80% | 1 | 2 |
| Foundation issues | 9 | 78% | 70% | 2 | 3 |
| Rental restrictions | 9 | 78% | 58% | 2 | 5 |
| Undivided co-ownership | 7 | 86% | 50% | 1 | 6 |
| Fire or flood damage | 7 | 71% | 100% | 2 | 0 |
| Water or mould issues | 6 | 100% | 86% | 0 | 1 |
| Liens or title issues | 6 | 67% | 80% | 2 | 1 |
| Auction | 6 | 50% | 60% | 3 | 2 |
| Oil tank | 5 | 100% | 71% | 0 | 2 |
| Heritage designation | 5 | 80% | 67% | 1 | 2 |
| Special assessment | 4 | 75% | 100% | 1 | 0 |
| Problem plumbing | 3 | 100% | 60% | 0 | 2 |
| Below-market rents | 3 | 67% | 33% | 1 | 4 |
| Flood zone | 2 | 100% | 50% | 0 | 2 |
| Court or receiver sale | 2 | 100% | 33% | 0 | 4 |
| Old or risky wiring | 1 | 0% | – | 1 | 0 |
No spin
What we still get wrong
We would rather you hear it from us. These are the weakest spots in the current release, and what they mean for you.
Value-Add is tagged too eagerly
Only 78% of our Value-Add tags are confirmed. Listings that mention renovations or upside often read as Value-Add when the reviewer disagreed. Treat it as "worth a look", not a verdict.
"Short-term rental allowed" is over-read
56% of our "Short-term rental allowed" signals are confirmed. Overall, about 2 in 10 signals we show are ones a reviewer wouldn't have added; it's the price of finding 88% of them.
Bathrooms are the hardest count
Half baths, ensuites and "3+1" style counts are written dozens of ways. We get 91% right; check the photos before you rely on it.
We only know what the listing says
Tags come from the listing's own words. If a seller leaves out the basement suite or the leaky roof, so do we. A tag is a lead to verify, not an inspection.
Getting better
Every release, measured the same way
Every release is re-scored on all the hand-checked listings we had at the time: 159 at the first, 1,388 at the latest, the same pool as the numbers above. We chose to find more strategies even when that adds a few extra tags: found went from 76% to 90%.
Strategies foundStrategies confirmed
100%
90%
80%
70%
60%
90% found
84% confirmed
Releases, Oct 10 → Oct 29, 2026. Property type stayed between 93% and 95% throughout.
Show as a table
| Release | Date | Listings | Property type | Strategies found | Strategies confirmed |
|---|---|---|---|---|---|
| 1 | Oct 10, 2026 | 159 | 94% | 76% | 82% |
| 2 | Oct 10, 2026 | 255 | 93% | 79% | 84% |
| 3 | Oct 10, 2026 | 255 | 93% | 80% | 80% |
| 4 | Oct 12, 2026 | 353 | 93% | 84% | 77% |
| 5 | Oct 12, 2026 | 498 | 94% | 88% | 78% |
| 6 | Oct 12, 2026 | 498 | 94% | 93% | 78% |
| 7 | Oct 12, 2026 | 651 | 94% | 93% | 77% |
| 8 | Oct 14, 2026 | 897 | 93% | 92% | 79% |
| 9 | Oct 14, 2026 | 897 | 93% | 92% | 79% |
| 10 | Oct 14, 2026 | 1,388 | 93% | 91% | 81% |
| 11 | Oct 18, 2026 | 1,388 | 93% | 91% | 81% |
| 12 | Oct 18, 2026 | 1,388 | 93% | 91% | 82% |
| 13 | Oct 21, 2026 | 1,388 | 93% | 91% | 82% |
| 14 | Oct 23, 2026 | 1,388 | 93% | 91% | 82% |
| 15 | Oct 23, 2026 | 1,388 | 93% | 91% | 82% |
| 16 | Oct 27, 2026 | 1,388 | 95% | 90% | 84% |
| 17 | Oct 29, 2026 | 1,388 | 95% | 90% | 84% |
| 18 | Oct 29, 2026 | 1,388 | 95% | 90% | 84% |
Method
How we test
The same approach researchers use to grade a classifier, kept simple and repeatable.
1
Pick real listings
We sample listings from our own feed, including rare types like cottages, land and fix-and-flips, so every tag has enough examples to grade.
2
Label them by hand
Each listing is read and labelled against a written guide: its type, every strategy it supports, its stated room counts and its deal signals.
3
Grade on unseen listings
We tune our rules on separate practice listings. The scores on this page come only from the 1388 listings that were kept aside, so they reflect listings the tags have never seen.
Our house rules
A few strategies follow fixed rules, applied the same way to every listing and to the hand labels:
- A duplex, triplex or other plex is a House Hacking candidate.
- A separate entrance makes a property a House Hacking candidate.
- Waterfront, lake, beach and ski-resort properties are Short-Term Rental candidates.
How the tags are made
Every new listing goes through the same steps, in seconds:
- ReadA small AI language model reads the title and description and answers a fixed set of questions about the property.
- CheckOur keyword rules back up or correct its answers, and catch strategies it missed.
- CountBeds, baths and garages come from the listing's own words; a count the seller didn't state stays blank.
- ProveA deal signal is kept only when the listing's words back it up.
Estimates
Value, rent and rental estimates
Some listing pages show an estimated value, long-term rent, local market medians and short-term rental potential. They are estimates built from public records, recent sales and other listings, never appraisals.
Where they come from
In Canada and the United States, value and long-term rent estimates come first from a statistical model of public records and recent sales; where it has no estimate, a property assessment from the last two years can stand in for the value. In the US, a third-party automated valuation model that compares the property with nearby listings answers next, and gives the ZIP code market figures. When neither has an answer, the value is the median asking price per square foot (or, without a size, the median asking price for the same bedroom count) of at least 8 listings of the same type in the same city from the last 12 months. Short-term rental potential comes from public short-term rental data: at least 5 entire-home rentals within 1.5 km that were booked in the last 12 months.
Read the range, not the number
Every estimate shows a range and what it is based on, such as "based on 12 listings within 1.5 km". Our own ranges run from the 25th to the 75th percentile of the comparables; a modelled value gets 5% either side (10% for rent, 15% for an assessment), and the automated valuation model gives its own. Comparables are asking prices, not closed sales, and asking prices usually run above what properties sell for. Signed-in users also see the point estimate and the comparables.
What they are not
Not an appraisal, an inspection or a valuation: nobody has seen the property, so its condition, renovations, a basement suite or a leaky roof are not in the numbers. Short-term rental revenue is modelled from reviews (half of stays reviewed, stays of at least 3 nights, at most 70% of the year booked), and local rules may not allow short-term rentals at all.
How often they change
New listings are estimated when they are found. Market medians from our own listings are recalculated nightly, estimates are refreshed after 7 to 90 days depending on the source, and short-term rental data is updated each quarter when a new snapshot is published. Where there isn't enough data, the page shows nothing rather than a guess.
Deal score
Deal score and analysis
Signed-in users see a deal score out of 10 on many listings, with who the deal suits, short notes on the card and a written analysis. Here is how often they agree with what the listing says, tested on 48 live listings that people checked by hand.
92%
Condition
Read right on 44 of 48 listings: renovated, needs work, a major project or no condition stated.
A rehab cost added only when the listing calls for work: 96% (42 of 44)
88%
Who it suits
Agrees with the labels on 42 of 48 listings: first-timer friendly, some experience or experienced investor.
Three levels, judged from the listing alone
83%
Facts in the listing text
We pick up 34 of 41 facts the listings state, such as fees, taxes, rent, financing terms and the number of units.
Only facts the seller wrote in the listing
92%
Claims backed by the listing
302 of 329 chips, figures and score reasons are backed by the listing, its estimates or the math on them.
Every claim on the card and the listing page
How we test it
1
Pick live listings at random
We sample 48 live listings at random, so the mix of property types, prices and places is the one you see on the site.
2
Label them by hand
Each listing is labelled by hand against a written guide, from the listing alone and without seeing our score or analysis: its condition, whether it needs work, who it suits and every fact it states.
3
Check every claim
We compare the score with those labels and check each chip, figure and score reason against the listing. These numbers are from November 1, 2026 and tied to the version of the score you see; we republish them whenever the score changes.
No spin
What the deal score still gets wrong
The score is a quick first read, not a verdict. These are its weakest spots in the current version, and what they mean for you.
Thin area data
Where few similar listings are nearby, the price comparison rests on very little, and the score can look better or worse than the deal is. Check what each figure is based on before you rely on it.
Unusual property types
Manufactured homes, land and homes sold without the lot don't fit the usual numbers. Lot rent, a land price that isn't included or a home built to order can change the cash you need and who the deal suits.
Terms written only in the text
Seller financing, a fee that already covers taxes, lot rent or a repair estimate is sometimes stated only in the description, and we don't always pick it up. When we miss one, the cash and monthly cost figures can be off.
Condition wording is easy to miss
Sellers describe a renovation or needed work in many ways and languages, and we don't catch them all. The condition we show is the seller's word, not an inspection; check the photos.
Questions
Frequently asked
What do "found" and "confirmed" mean?
Found (recall) is the share of the tags a listing really deserves that we show. Confirmed (precision) is the share of the tags we show that a human reviewer agreed with. A tool that tagged everything would find 100% but confirm very little; one that tagged nothing would never be wrong but would find nothing. We report both so you can see the trade-off.
Why do you favour finding over confirming?
Investors told us a missed House Hacking or As-Is tag costs them a deal, while an extra tag costs a few seconds to dismiss. So our rules lean toward showing a possible strategy or signal for you to check, and that is why found is higher than confirmed.
Do you guess a missing bed or bath count?
No. If a listing does not say how many bedrooms, bathrooms or garage spaces it has, we leave the field blank rather than guess. Blank is better than wrong when you are filtering for deals.
Which listings were used for this test?
Real listings from our own feed, 1388 for property type and strategies and 1310 for deal signals, each checked by hand against a written labelling guide. None of them were used to build or tune the rules, so the scores show how the tags do on listings they have never seen.
How often is this page updated?
The accuracy numbers are republished every time we change how listings are tagged; the last update was October 29, 2026. The live counts at the top of the page refresh every 10 minutes.
Does a tag mean a property is a good deal?
No. Tags describe what the listing text says, such as "sold as-is" or "separate entrance". They help you find candidates faster; always do your own due diligence, inspection and numbers before making an offer.
I found a wrong tag. Can I report it?
Yes, please do. Send us the listing link through the contact page. Wrong tags are exactly what we use to improve the next release. Contact us.
Put the tags to work
Filter 3,090 live listings by strategy, deal signal and property type, then let new matches come to you.