Here’s What Artificial Intelligence Tells Us About Toronto Real Estate Prices

We don’t just deliver real estate news at Better Dwelling, we’re trying to predict the next story before it happens. So we’ve recruited a few helpers in our arsenal, and thought we’d introduce you to our newest intern – IBM’s Watson. For those of you that don’t know who Watson is, he’s an artificial intelligence (AI) engine that has beat humans at Jeopardy, is working on curing cancer, and one day he’s going to help you buy a home. He’s got a pretty pimp resume.

One of the tasks we’ve assigned him is to find out what the primary drivers of average prices are in Toronto. So we loaded the past few years of sales data from the Toronto Real Estate Board (TREB), and asked him what he sees. We’re saving his predictions for another day, but we thought we would give you a preview of what he’s observed in buyer behavior.

What Drives Average Prices?

Watson believes there are 3 primary drivers of price, and no it’s not location, location, location. Total listings available in relation to days on market (DOM), DOM in relation to new listings, and DOM in relation to sales were the most important correlations. According to the artificial intelligence engine, each scored a 94% relevance in influencing average prices.

DOM In Relation to Total Listings Available

The relationship between DOM and total listings available have one of the highest correlations to price. Less days on market, and less listings equals higher average prices. He’s not blowing us away with this insight, but it’s interesting that a machine learned that in less time than it took you to read this sentence.

DOM In Relation To New Listings

DOM in relation to new listings was another important driver. Less days on market, and less new listings meant higher average prices. Once again, another supply and demand relationship observed, and confirmed.

DOM In Relation To Sales

Fewer DOM and fewer sales is an interesting correlation. A bit of a curveball, since you might assume fewer DOM and more sales would be the driver. He may have just observed that fewer sales and fewer DOM means people are more focused on fewer properties, regardless of the size of inventory.

This makes sense if you think about it, since people might be competing over one house, regardless of whether or not there is more inventory. It breaks the generic rule of less inventory and higher prices. This was one of the more interesting ones in our opinion.

What Does NOT Drive Prices?

Apparently, everything else had little correlation in Watson’s opinion. That isn’t to say other factors are irrelevant, but he is saying that when building a predictive pricing model, we should prioritize these factors above all. This includes DOM when not compared to the three metrics above. A decline in DOM did not always show increasing prices. So listing your home with an offer by date in 10 days, in say… 1990 might not have created the price increase you might expect.

These insights aren’t exactly groundbreaking, but keep in mind a machine extracted them. It’s important to note that these aren’t observations of the mechanics of real estate, but these are observations of the behavior of buyers. The important difference being buyers responded to these factors, they didn’t necessarily create them.

Want to keep posted on the latest in real estate news? Like us on Facebook for insights right in your feed.