Quantarium is an unusual company in the fintech space. Its founders, who have deeply technical backgrounds in computer science and quantitative genetics, weren’t seeking to disrupt a traditionally manual industry with artificial intelligence and machine learning, although that’s what happened. In fact, Clement Ifrim and John Smintina weren’t trying to start a company at all.
Instead, the friends were comparing notes as both began considering investment properties in the Seattle area and became intrigued by the calculation of home values.
What were the critical factors determining home value, and how could you apply a consistent strategy within the context of frequent market volatility with so many variable features from one house and neighborhood to the next?
Was there a model that could predict which properties would make better investments than others?
Ifrim, a former computing research scientist and Microsoft alumnus, together with Smintina, who holds a Ph.D. in quantitative genetics and has a vast experience in data warehousing, started building innovative models, leveraging their collective expertise to analyze whether they could gain more reliable insights for addressing and resolving the multi-layered valuation challenge.
Their mutual fascination with applied artificial intelligence and genetic data modeling led them to define and deploy sophisticated machine learning techniques and even develop a prototype to rank properties according to their investment potential.
After an extended period of design, development, testing and evaluation, they realized the technology they had created for their own use had broader market applications and they began commercial considerations for Quantarium in 2013.
The name Quantarium, relying on foundational concepts in quantum physics, reflects the company’s focus on leveraging in an initial non-discriminatory manner, all possible data, up to the smallest pieces of information available, as the starting point for its innovative products and services.
Both Ifrim and Smintina graduated from the highly regarded “Liceul Informatica Iasi,” a computer science school in Romania, before going on to get advanced scientific and technology degrees and pursuing careers in the global technology market.
Shortly after founding Quantarium, they reached out to another illustrious alumnus, Mihai Petriuc, bringing expertise in information retrieval and evolutionary computation, to serve as the chief software architect, and rounded out the leadership team with Malcolm Cannon, a veteran enterprise technology executive.
From its earliest days, Quantarium attracted a brain trust of AI scientists, ML experts, researchers with advanced degrees in mathematics, and enterprise software engineers who are now leveraging their expertise in artificial intelligence and machine learning to create compelling solutions for companies in the real estate and mortgage industries.
These solutions are constantly evolving, delivering sophisticated intelligence for better, more accurate decision-making, from valuations to portfolio services and more.
QUANTARIUM VALUATION MODEL
While working on their ranking system, Ifrim and Smintina realized the rankings required an incredibly accurate valuation model, something with no more than a 5%
error rate. That level of accuracy wasn’t available outside of a very expensive commercial model, so the Quantarium team set out to design one specifically for the residential real estate market. The result, Quantarium Valuation Model or QVM, exceeded their expectations.
QVM relies on a cascade of proprietary solutions, including neural networks and genetic modeling that mimic biological processes, to get the most accurate results. This true machine-learning approach runs hundreds of thousands of generations of property specifics and values to understand and optimize information on micro-markets at the neighborhood, ZIP code or county level.
QVM learns and self-turns to what is valuable among its robust and ever-expanding set of property data, and factors in the effects of local events like foreclosures and recent sales. QVM is continuously evaluating and learning — a process that produces further data points which become part of any new property valuation.
However, the potential roadblock in any genetic modeling approach is how to deal with outliers that can skew overall results. Quantarium’s breakthrough in this area delivers incredible accuracy and is part of the company’s “secret sauce.”
“Artificial intelligence and machine learning aren’t just buzzwords for us — we are a true AI company. That’s one reason QVM is ranked as the No. 1 most accurate valuation model,” Ifrim said.