Using AI to Build SFR Empire – Data science helped Sean Dobson, CEO of Amherst Holdings, create a fortune in the real estate crash. He is now deploying A.I. to acquire properties building a SFR empire.
America’s A.I. Landlord.
The increase of the single-family rent sector represents deep changes in America’s families ‘ finances and attitudes. For many middle-income workers, home ownership, long a cornerstone of economic stability, has become unattainable or undesirable— for reasons including tighter lending norms, high college debt burdens, and lagging salary development and savings. According to Yar-deni Research, slightly more than one out of three families who would have purchased their first homes before the financial crisis are now either renting or still living with their relatives.
Dobson established the precursor to Amherst in 1994, and by the early 2000s, Amherst was selling pension funds and insurers $25 billion a year in MBSs. During the housing bubble, when his models predicted a disaster in “Alt-A securities,” loan packages granted to homeowners who had often refinanced multiple times some with LA Hard Money Loans, the seeds of his big score were planted. “The market predicted a 5 percent default rate, and our models showed that it would be 30 percent[ even] if house prices did not drop at all,” Dobson recalls. He hired a group of investors who took short positions in Alt-A, making a profit of $10 billion—10 times the investment, according to Dobson — when home prices fell.
The front-row seat of Dobson at the collapse of the homes helped him acknowledge the rental chance. He had started a campaign by 2011 to convince investors to finance a fresh venture— a fund to purchase and lease industrial-scale single-family homes. The potential was seen by some of his former partners. “Rentals for single families are basically a large data game,” says Mariner Investment Group head Curtis Arledge. “If you purchase on a scale, you gather all types of data. That information provides him with a competitive advantage.
Based on studies such as Negri’s, Amherst now objectives in 30 metro regions around 1,000 zip codes. Home selection is the task of the extremely automated buying scheme of Amherst. A dozen purchasing experts monitor leads on their workstations in their 19th-floor office on Madison Avenue in New York City, provided by a proprietary program called Explorer, an offshoot of Dobson software created for pricing mortgages. The team receives alerts every morning about freshly listed homes that fulfill their price variety and geographic criteria— about 1,400 listings a day.
Explorer also performs a separate calculation, finding three rented homes within a two-mile radius that are close to the newly listed home in terms of age, size and bed and bath specs. Machine learning helps the software estimate what each house would rent for on the basis of these “comps.” Explorer then churns out an estimated “rental yield”— net rent after costs such as taxes and maintenance, divided by all – in costs.
If that yield meets the goal of Amherst (which Fortune estimates are between 5% and 6%), the team will create an offer. Approximately 20 percent of the listings of each day qualify; Amherst bids these applicants no more than 12 hours after the first listing, making all-cash deals. Approximately 10 percent of their offers— about 30 homes a day — are accepted and contracted. Amherst dispatches inspectors during the grace period to evaluate the situation of each home. The buildings pass muster and join the portfolio of Main Street Renewal unless they discover deadly surprises such as a cracked foundation.