Two Sigma has built a data-driven apparatus to help it assess prospective commercial real estate investments.

The firm’s so-called market selection tool—one completed application among a suite of tools currently in the works—melds traditional with alternative data to find the best regions in which to invest.

“We can look at markets and investments and see what is the relative risk of suburban Denver versus CBD [Central Business District] Houston,” said Drew Conway, head of data science for Two Sigma Real Estate, a new division of the firm launched in April.

Conway was speaking at the Ai4 2021 conference on August 19.

The new tool draws on alternative data ranging from census information to cell phone geolocation ‘pings’ to understand real estate supply and demand—the latter showing how fast people are returning to city centres as the Covid pandemic recedes. The tool scores the relative risk of different geographical markets in which Two Sigma might consider investments.

• READ MORE: The value of real estate markets dwarfs other asset classes, but a lack of data has hindered its development. Click here to read more.

The application is part of a broader drive at the New York City-based hedge fund to open new lines of investing in areas seen as data rich. In 2019, it set up a team to invest in private equity, around the same time as work began on its real estate project.

“There’s lots of capital flowing in and out of the commercial real estate business, and a large scale of data being generated,” Conway said. “When we see this scale of industry, we want to look for variations in that data to create data-driven [investment] strategies.”

The commercial real estate business generates information all along the investment chain, Conway said—from when a broker first lists a property, to buyers’ and sellers’ bids and offers, to commercial loan information. Once an asset is owned, it becomes a “data generating machine”, throwing off information about property upgrades, new construction, tenancy occupation levels and rental payments.

Two Sigma aims to combine traditional data on transaction prices, rents, lease terms, property characteristics, debt levels, tax information, zoning permits and so on, with alternative data such as foot traffic density, consumer spending patterns in local areas, key public announcements, and satellite imagery on construction activity. Conway said the objective was to build a consolidated “golden core” of information.

We can look at markets and investments and see what is the relative risk of suburban Denver versus Central Business District Houston

Drew Conway, Two Sigma Real Estate

Such data can signal whether an asset is going to be sold, he said, or help identify property layouts and style features that are driving higher values. It can predict the effect on prices of policy changes such as zoning adjustments or tax changes.

“We are observing lots of micro behavior from consumers and businesses, and macro behavior on where markets are going [and which] markets are growing. If we understand where liquidity will be, we can identify assets that may be selling in future.”

Both old and new data has strengths and weaknesses, Conway says. “Commercial real estate data is prone to noise. Proprietary alternative data is faster-moving but difficult to work with.”

He pointed to consumer spending patterns as an example. “We can get [the data] quickly, but the question is how to fold that with traditional data to make it make sense.”

Two Sigma’s push into real estate matches similar moves by other investors to apply quant techniques in private markets.

Private equity houses such as KKR, NBPE and Warburg Pincus have adopted quant approaches to investing, leveraging proprietary data to identify prospects and gauge risk. In January, AQR Capital Management’s Cliff Asness stated that the firm was “musing” about the potential of moving into private equity.