Real Estate fello

Enhanced AVM
FoxyAI fello

An Enhanced AVM Case Study

Outcome

“The FoxyAI score significantly improved our Automated Valuation Model (“AVM”), giving a 5% improvement in PPE10 (percentage of time the AVM is within 10% of the benchmark) and a 2% improvement in MdAPE (Median Absolute Percentage Error).”

– Head of Data Science at a Large Regional iBuyer

Situation

An Instant Buyer (“iBuyer”) is a real estate company that uses algorithms and technology to buy and resell homes quickly.

In 2021, a Large Regional iBuyer that successfully bought residential homes entirely online decided to expand its offering to brokers and agents, providing them with the tools to offer their own iBuying solution. In creating this new offering, the Large Regional iBuyer’s leadership identified the key components necessary to execute and become more profitable. One essential component was to source the technology to create a hyper-sophisticated AVM to give its team and new B2B audience the confidence to acquire properties instantly.

Challenge

It is a data science and artificial intelligence (“AI”) feat to improve a successful iBuyer’s proven AVM. The iBuyer must have created an accurate model to become successful, and to improve upon it takes the latest in AI advancements and a discerning data scientist.

This Large Regional iBuyer faced an added challenge of entering the saturated B2B market. Now, in addition to improving a demonstrated AVM, the iBuyer’s team had to attract an astute agent and broker audience and needed to find the right technology partner to help them do so.

Business Benefit

After vetting technology partners, this Large Regional iBuyer chose to run a pilot AVM test with FoxyAI’s Models. The iBuyer’s data science team wished to incorporate FoxyAI’s Condition Score Model into its current AVM to see if it would improve its accuracy.

The FoxyAI Condition Score is a proprietary model that analyzes a company or government entity’s property media and provides a continuous condition score, using a 6-point scale ranging from Brand New to Heavy Damage/Not Livable. It is based on the scoring system from the Uniform Appraisal Dataset used by Fannie Mae, Freddie Mac, and others for underwriting.

The iBuyer’s pilot test ran 19,105 properties through FoxyAI’s Condition Score Model, totaling to approximately 400,000 photos. The iBuyer’s team then used the scores to retrain its AVM model and recompute the values. The iBuyer’s FoxyAI Enhanced AVM “significantly improved” its original model, reporting a” 5% improvement in PPE10 and a 2% improvement in MdAPE.” In dollar terms, this AVM enhancement on average brought each of these 19,105 properties ~$3,000 ($2965.52) closer to the sales price, accounting for a ~$57,000,000 improvement in property value.

Today, this Large Regional iBuyer is rolling out its AVM which has been enhanced with FoxyAI’s Condition Score. The iBuyer team successfully improved its demonstrated AVM in order to expand its product offering to a new market. In addition, the iBuyer increased its brand footprint and industry recognition while adopting the latest in visual intelligence technology.