GTA home values, HouseSigma App, Accuracy of Automated Value Machines, Predictive Sale Prices
By Sue Anfang, February 29, 2024
HouseSigma is a popular app for home buyers looking to predict sale prices. In fact it’s so easy to use that I use it too! In the US they use a similar app you may have heard of called Zillow.
Another feature that’s valuable is you can look up sold prices for a home and easily check the listing history to see what the home previously sold for, if the house has been re-listed recently and how long it’s been on the market. And it will show you lots of other good stuff like neighbourhood values, too.
Of course – it’s all linked to their HouseSigma agents should you want to book a showing. Which means HouseSigma is also a Real Estate Brokerage using this tool to get leads for their company.
So how accurate is House Sigma anyway? It’s a question I’ve been wondering. So I tracked a bit more than a dozen listings over the past couple of months and watched them sell so I could see for myself. And you might find the answer surprising!
I chose a variety of listings in the GTA, Markham, Mississauga & Durham for my sample size.
And I also threw in a couple of homes in my neighborhood to check Scarborough house prices that I knew very well.
If you go onto HouseSigma, what do you see?
Here's are two examples of how the predictive pricing looks on a listing display - I took out the address for privacy, but added actual sale prices (in the red bar)
The listings are quite attractively displayed, and include lots of property information, all the photos as well as local real estate statistics. I don't believe they cover every market in Ontario.
First, what does HouseSigma say about themselves?
“HouseSigma is a leading technology platform that utilizes artificial intelligence technology to estimate Canadian home values in real time. With one click, home buyers can obtain an accurate automated home valuation for every listing within seconds. The HouseSigma algorithm also correctly identifies similar nearby sold properties to help buyers determine their final offer price!”
And further, on their FAQ page, they answer the question: “why are some of the estimated values not correct?”:
“The estimated price should be used as a starting point for your research rather than a number that you should follow to the T. There is no perfect AVM (automated valuation machine) in the world, including HouseSigma’s. How our algorithm normally come up with the number is based on long term market trend and nearby comparable. Click “Contact Agent” in any HouseSigma property listing page to start a conversation with a HouseSigma Licensed Representative and get more insights on up to date market analysis”
One thing you may notice is their automated valuation machine algorithm changes day to day. Just for fun sometimes I will go in and check their predictive price on a listing – and presto change-o it will be different on a Wednesday than it was on Tuesday.
So what did my very scientific survey show? A HUGE variation in sale vs. list price!
So why is it so wrong? And why is it so unpredictably wrong?
Because it is based on an algorithm (and have signal is not specific about what that algorithm includes), it does not take into account many factors that affect price.
HouseSigma does not seem to be able to see the difference between a renovated house and a house in original condition.
But the biggest factor is under-pricing. Many seller use under pricing to get a bidding war. Similar properties can use very different starting prices & eventually end up around the same sale price. Occasionally, (as in the case of one of the houses here) you’ll see a property underpriced $50,000 or more less than similar homes that were previously underpriced.
Or in the case of one house that was in the news earlier this year it had been extraordinarily underpriced to the tune of the starting price five years earlier! So that predictive price was way off what it sold for.
Conversely, if all the homes in the neighbourhood are under priced & sell over asking, HouseSigma will assume all homes sell for over asking in that area. So something priced at market value have a higher predictive sale price. Again, one of the houses on the chart fell into that category - and sold lower than predicted.
A cautionary tale for sellers
And for that particular house - I had valued it in the summer - comes a cautionary tale. I had valued it for higher than the list price above in the summer. I used the technique below. It was a very unusual house, and the features that made it unique were not necessarily features buyers value, or will pay extra for. But they chose to list with an agent that listed it way higher than my valuation. In fact, they told me all the realtors they spoke to valued it higher then me (did you know some realtors will over-value a home to secure a listing?). So what happened? The house sat for months at the high price in a slow market. Eventually after several price reductions, it sold for - guess what - about $100,000 less than my original valuation.
So then what’s the best way to figure out market value? In most cases, a realtor will look at individual homes with similar features that have sold recently close by. Then I will figure out a potential sale range.
Curious how I approach valuing a home? Give me a call at 416-562-5002
As a consumer can you check sold prices to do your own research?
Yes you can! As a matter of fact my website comes with 2 great features you can use for your own research.
1. GPS enabled App - allows you to view homes for sale and recently sold & save home seearches
2. Automatic emails of sold properties - I can set that up for you to keep track of a specific neighbourhood - it's no obligation!
If you want your own personalized app; email me at: [email protected]
