CRE Giant Bgo uses AI to find underestimated property in unlikely areas

The version of this Article first appeared in CNBC Nevset Play Nevsletter with Diana Olick. Playing property covers new and developing opportunities for real estate investor, from individuals to the risk of capitalists, private capital, family offices, institutional investors and large public companies. Apply To receive future releases, directly to your mail.

John Carrafiell, Ko-director of BGOG, a global real estate investment manager with 89 billion funds in assets under management, occupies great pride in the fact that he sits right next to his main scientist.

The investment strategy, regardless of the market, always relied on research and data, but artificial intelligence took over to a completely new level, transforming investment research models only a few years ago and putting them on steroids.

Carrafiell, who was in real estate work in approximately 40 years, said that it was increasingly frustrated by research and methodology in the sector, which he said that she had not really changed at all times. Everyone seems to look at the same information and give themselves to the same conclusions. The question he said he was constantly asking it was, “How do we really surpass ourselves?”

The answer, found, was to analyze all past offers of his companies that return for 20 years, using only a computer model and extract the human element from it. What found the model that the local market was fully determined by the performance or insufficient performance elected in the investment.

It can sound triten – considering that the property of the property was always “location, location, location” – but the results said that its team was almost entirely focused on the basis of the local market when choosing their future investments, not so much on property prices.

There are, of course, research firms that analyze and rank local real estate markets, but Bgo revealed that their results are a bit random, according to Carrafiel. Instead, she looked in her past and built a model that exactly asked what she took the best and worst performance. The model includes all kinds of local market data, including demographic and supplied trends unique to each location. AI was then given that model increased data volume and speed.

“We took thousands of data entries, many who are released from the Government, many, for example, telecom providers, excellent data. We found the key,” Carrafiell said. “And we know it’s true because it hindered him.”

Bgo used his science on data to inform the decision on the role in industrial development in Las Vegas with a partner in the North Minister. Other data models suggested that it was not especially good investment.

Carrafiell said the “best research” indicated that the investment would be mediocre in terms of performance and return.

“But our model was screamed, they will explode. We went out for rent for $ 5.88. We got rent in a range of $ 9 per square,” he said. “It doesn’t happen in commercial real estate. It’s not happiness.”

The model, he explained, saw that in the interior of California became too expensive, and then analyzed logistical routes. It revealed that companies could save whether they were in Las Vegas instead, where rentals, taxes and work are cheaper.

“So, you had an extra two-hour drive, but you saved as 60% on your total cost and it was a video model,” Carrafiell said. “The tenants we have there we serve the entire region. They are not served by Las Vegas.”

Bgo ran into a similar analysis for investment in Florida and rust belts, resulting in a great return on its investment.

“We think our performance has significantly increased as a result of this model,” Carrafiell said.

But he admitted that although the accuracy of the model is to improve dramatically, hypothesizing, “Boeing can be evicted from Seattl, and the model cannot predict it, can’t he have an idiosyncratic stuff.”

While the BGO investment is focused on upside models for potential properties, his lending team looks at modeling for the decline, because it lies its risk.

New ITERATIONS The research model down the road will include the allocation of property in different sectors of commercial real estate. The model would ideally propose the optimal portfolio mixture. The possibilities are still growing, which is why Carrafiell says it is sent to the data like never before.

“Ai is a booster and an accelerator that allows us much more, but it’s really a science of data,” he said. “It is (like) six-bed, dedicated scientific data teams sitting next to your General Manager and in addition to the asset and acquisition management team.”