How Proportunity uses machine learning to detect bargain homes | Industry

Breaking Tech news from the top sources


Budding homebuyers struggling to get their feet on the property ladder can now get an automated leg-up from .

The London-based startup unearths hidden bargains on the market by using hidden gemes out of piles of data. It then offers loans based on the predicted future value.

© Proportunity
© Proportunity

The datasets they analyse cover everything from crime rates down to whether a Whole Foods is opening in the neighbourhood. There are around 100 different indicators in total, with half of them covering the area-specific factors that reveal supply and demand such as unemployment, transportation links and Ofsted school ratings. The other half focus on the property itself, to understand how it compares to other on the market. These factors include the property’s size in square metres, its number of bedrooms, and its energy consumption.

The historical impact that these factors had in other geographies helps Proportunity understand the effect that they will have on the future price.

“We analyse all transactions since 1995 with most of those descriptors, and then we learn from similar areas to forecast areas and houses that we believe are underpriced,” Proportunity CEO and cofounder Vadim Toader tells Techworld in the Google Campus London in Old Street, where his startup recently completed a machine learning residency.

The indicators can reveal some surprising investment opportunities. Analysis of police arrests and the chemical compounds in sewers that people flush down their drains shows that when the use of crack cocaine drops gentrification could soon arrive, but when the crack is replaced by cocaine gentrification may already be complete.    

If someone buys a home that Proportunity recommends, the company will offer them an equity loan of up to 15 percent of the price.

The loans they provide differ from those offered by traditional mortgage providers, which charge an interest rate that reflects the risk of the loan. This risk is determined by assessing the applicant’s income, expenses and credit history.

“The problem is when they determine how risky a loan is, mortgage lenders tend to only look at the borrower,” says Toader. “For example, if you were to buy a farm in Northern Ireland that will never grow in price because there’s just no demand in that part of the country or a house in a gentrifying area in London for the same price, the mortgage lender would always charge you exactly the same rate. What we do, is we estimate the credit riskiness by also looking at the asset.”

If they predict that the house in London will have a bigger increase in price than the farm in Northern Ireland, Proportunity will offer the borrower a lower interest rate on the loan.

“That’s the fundamental difference,” Toader adds. “What’s cool is because we add a bit more, then suddenly the max mortgage you can take from your bank becomes cheaper per month, because they see it as less risky since there’s a bigger buffer between the price of the house and their share. So you can buy a 15 percent bigger home with us than you could without us instantly, which tends to make people quite happy, because then they can buy another bedroom, or they can move slightly closer to the city.”

Building a property business 

Nowhere in the UK is the housing crisis more acute than in the Romania-born Toader’s adopted home city of London, but the idea for Proportunity sprung from similar struggles across the Atlantic.

Toader’s wife was looking to buy a home in New York, but as every real estate agent claimed that they covered the top area for investment it was difficult to know which locations offered the best future value.

The couple studied the market and spoke to analysts and friends and eventually found a big home that was well-connected to the city centre.

“Then it grew in price by 60 percent in eight months,” Toader remembers. “We could have seen that, because Whole Foods had just opened up, you had a lot of new construction in the area, Zuckerberg had donated $150 million to improve schools in that area. All the typical things you look for when you’re buying a house had changed, but no one was tracking that.”

Toader quit his job at an investment firm to turn his idea into a business. He and Proportunity cofounder Stefan Adrian joined in the Entrepreneur First accelerator and then entered Google Campus London’s six-month machine learning residency to refine their concept and bring a product to market.

They initially planned to provide loans by collaborating with the big mortgage lenders but that idea quickly changed when they realised they could do the job themselves.

In September 2017, they headed to TechCrunch Disrupt in San Francisco as part of a group of startups selected by the UK government to represent British innovation. 

We met with a few VCs when we were there, and they pushed us and said that our business model isn’t as scalable as they would like, and we agreed,” he says. “At that point we pivoted. We did forecasting in real estate – basically forecasting house prices – and the pivot went from selling information to the banks to becoming a bank, because what we were finding out was that the banks were moving too slow.”

The sector’s biggest event, the UK Finance Annual Mortgage Conference, was approaching. The main sponsors had just dropped out, and Proportunity stepped in to replace them. The team worked with Google Campus’ roster of experts to plan a presentation for the conference that attracted the attention of an FCA-authorised mortgage lender who saw a new opportunity for his business.

“He had the knowledge, he had the FCA authorisation, and he wanted to do it somewhat differently, so we said let’s merge teams and do it together,” Toader recalls. “That allowed us to become, I think the fastest ever FCA-authorised mortgage lender. Instead of a year and a half, it took us about three months.”

Next steps on the ladder

Machine learning can anticipate changes in prices before the market can react. This offers an opportunity for data-savvy buyers, but they will still struggle to compete with the machines.

“There’s usually about a five-year lag,” says Toader. “So try to look at statistics about that, and see where you see a massive decrease in crime, but the price is still flat, because that means there’s a trend moving up. It’s the same with transportation, it’s the same with schools. Just do a few areas, and just pick out the best ones you find. But it’s probably easier if you go to our website.”

Toader and his team are now looking for external customers and aim to help 50 people buy their first home by the end of the year.

They’re targetting first-time buyers as they have the biggest issues with deposits because their salaries have stayed flat while property prices have exploded.

The first of these buyers that Proportunity helped was an employee. When he bought a property, Toader brought a bottle of champagne to toast the milestone. Before long he too was using the service to buy his own place in London, and soon four core members of the team were living within 200 metres of each other in the same underpriced area.

“We’re now calling ourselves the first mortgage lender built by first time buyers, for first time buyers,” says Toader.

Advertisements

You might also like More from author

Comments are closed.