In the late 1990’s a new type of person started to pop up in stock market trading floors. These traders did not have a background in finance, as was the norm, but instead had PhDs in fields like mathematics and statistics. They were also not very well received by the financial establishment, they were referred to by the slur “quant.” Eventually, the mathematicians started outperforming their peers and the term quant was embraced. Now traders promote their use of advanced computing by calling themselves quants.
Quant is short for quantitative analytics and it is a way to use statistical analysis on large amounts of data to identify trends and correlations. It has become the go to way for traders to understand what the probabilities are for different macroeconomic trends and asset price movements. But despite its prevalence in equity and options trading, it is still not widely used by many in the real estate industry. That might be because many real estate firms don’t have teams with a background in mathematics and the data infrastructure needed to apply the same techniques. “Most companies don’t have the resources for advanced analytics,” said Josh Panknin, Director of Real Estate AI Research & Innovation at Columbia University Engineering.
So to help the real estate industry adopt a more quantitative approach, Columbia is working directly with firms. “Companies come to us with problems or new capabilities they want to develop. Then we vet these ideas to see if it’s feasible to develop a data solution,” Panknin said. They use huge troves of data to help them understand how different markets might react differently to volatility and other economic or real estate market indicators.
Getting the real estate data to a point where it can even be used by a quantitative approach is the first task. “We often start with government data because it is more consistent across geographies, which makes it easier to compare,” Panknin said. “Data provided by local jurisdictions and firms is much more difficult to use because most of that data comes in different formats, types of information, amounts, and collection/distribution formats.”
One example of this is the data about the city of Dallas. Each one of Dallas’s counties provides different sets of property level data and each compiles that information differently. These differences might not seem like much but they make it extremely difficult to make the connections needed to provide useful analysis. What Panknin and his team have found is that with the right approach and focus on data engineering, they can make predictions about the future with pretty high accuracy. “We look 6 quarters out pretty accurately, after that it starts getting pretty hard to make any reliable predictions,” Panknin said.
An example of what his team has uncovered through their work is how certain markets perform during different phases of the real estate cycle. A key factor, it turns out, is geographic constraint. According to the analysis, cities that cannot expand outward, such as New York or San Francisco, experience more volatility compared to cities with ample surrounding space like Denver or Dallas. Geographically constrained cities tend to see larger price drops in down markets but perform better during bull markets than their unconstrained counterparts.
A major hurdle that Columbia and the rest of the academic world must navigate is the general public’s lack of understanding about what advanced computing like AI can accomplish. AI is fundamental to quantitative analysis but it is only the engine for extracting the insights, not the critical factor in whether or not a problem can even be solved. “Most people come to us just wanting ‘some AI’ to solve a problem. They expect data in and miracles out. But they don’t realize that only about 5% to 10% of developing AI is building models. The rest is structuring the problem and getting the data into the right format. If you don’t have the problem structured computationally and don’t have the data properly formatted, your models simply will not work,” Panknin said.
It might be a while before we see real estate investors calling themselves quants but the underlying principles of quantitative analysis are already being incorporated into some of the country’s largest firms investment theses. Much like advancements in science and mathematics, quantitative analysis will likely be spearheaded by the academic world, not the private sector. We might soon see more real estate companies partnering with universities to help them tap into the forefront of computer science in a way that will give them a bit more insight into understanding markets and predicting the future.
Chudo Slot Casino unexpectedly gifted me with a win of $1,780 on slot machines. My best win in my entire gaming career!