In this article, we’ll delve into the dynamic world of machine learning in real estate. From pinpointing lucrative investment regions to streamlining deal sourcing, we’ll explore how machine learning can empower you to stay ahead in this competitive field.
Recent data showcases a seismic shift toward machine learning adoption, with 44 percent of CRE investment firms actively integrating it into their strategies. If you’re a General Partner seeking the edge in this dynamic market, you’ll want to explore how machine learning is transforming the industry – and how you can get started with it in your own firm.
What is machine learning?
Machine learning, in its essence, is a subset of artificial intelligence (AI) that enables systems to learn and improve from experience without being explicitly programmed. Machine learning requires:
- High-quality data in vast quantities
- Cloud computing to store this data
- Algorithms or mathematical models to process and analyze the data
- Significant time and financial investment
- Human expertise
By leveraging machine learning to analyze vast quantities of granular and disparate data, real estate investors can unlock a myriad of valuable insights, from discovering high-potential investment regions to sourcing deals and providing more accurate asset valuations. We’ll explore these applications in greater detail in the sections to follow.
Notably, recent statistics affirm the upward trend in machine learning adoption within the real estate industry, with 44 percent of CRE investment firms surveyed in Altus’ 2022 global research report, ‘The State of Data Science in CRE Investing,’ indicating that they’re actively incorporating machine learning capabilities into their strategies.
How to Leverage Machine Learning in Real Estate
Identifying Regions to Invest in
Machine learning allows you to go from a submarket approach to an asset-based approach.
Here’s how most firms operate: they research a specific market and submarket, decide on a relevant submarket to invest in, and buy opportunities when they arise.
This approach is limiting: being too focused on specific submarkets often leads to the ostensibly less favorable ones getting ignored. However, even in submarkets deemed unattractive or lacking opportunities, so much is happening at the asset level.
That’s where machine learning can be a powerful tool.
Real estate firms that are using machine learning build systems that take various data points, including how much land is being developed in a certain submarket, property price growth rates, employment growth, population growth, and population demographics, to help them understand where the assets with the biggest potential upside are located.
Certain land investment firms, for instance, use machine learning to help them identify markets where prices are not only going up, but where substantial amounts of land are being turned into homes. Based on that, they decide on regions of investment – not the other way around.
For a land investment firm, this approach means they don’t get confined to a specific town or submarket – rather, they simply follow the data to the assets.
Bottom line, this enables them to to find assets that are located in the path of development on a 3-5 year time frame, and buy ahead of the substantial appreciation that occurs as the land becomes suitable for near-term development.
Tip: If you’re considering leveraging machine learning to identify opportunities, remember that the models don’t need to be perfect – they need to be good enough to guide you to the right starting point. Once you have a shortlist of assets to invest in, it’s all about your manual due diligence and judgment.
Sourcing Deals
Sourcing deals is another realm in which machine learning can shine for real estate firms. While scouting for opportunities manually is feasible when dealing with a handful of properties, machine learning becomes indispensable when the volume increases exponentially.
The scalability of technology allows us to efficiently sift through a vast array of listings and identify the most promising investments. One notable advantage is the ability to spot off-market opportunities, often concealed from the broader market, and approach landowners with a reasonable offer.
Machine learning algorithms can analyze a multitude of data sources, helping you uncover those hidden gems. The more you engage in off-market deals, the more valuable machine learning becomes, as it streamlines your search for untapped potential and identifies opportunities the human eye might miss.
Tip: There is always an opportunity cost. You could hire a team of analysts to source deals, instead of building an in-house machine learning tool. It ultimately depends on your budget and business goals: for firms looking to meaningfully expand, the technological investment and the scalability it provides make it a worthwhile investment. Worst case: the tech saves you from having a large junior team at your firm. Best case: the tech helps you make the best possible decisions.
Classifying and Valuing Land Assets
In contrast to other real estate asset classes, which can be fairly accurately valued using measurements like cap rates and cash flow, land that isn’t immediately ready to be developed is a trickier proposition.
With land, you can use comparables to make relative valuations or calculate potential earnings from developers further down the line, but these aren’t sufficiently accurate. That’s where machine learning can prove so powerful.
For example, a land investment firm could build a technology that automatically classifies undeveloped land assets into ranch, commercial, or single-family residential to allow for distinct valuation modeling and evaluation. A simplified example would be if the land is on a slope, it’d be classified as a potential ranch rather than a commercial property.
Data scientists can then feed the algorithm a historical data set of list prices of land, and an array of other data points, which it can use to forecast the discount to the list price that the property will be buyable at.
Advice for getting started with machine learning
Integrating machine learning into your investment firm is far from a low-hanging fruit, but when executed correctly it can reap major rewards.
Our main pieces of advice are: pay up for the best talent, use a recruiter, and ensure your tech and investment teams work together. Let’s unpack these:
Don’t Be Afraid to Invest
The challenge when hiring is that you’re competing with top tech companies for a small pool of highly specialized individuals.
However, it’s imperative not to compromise on talent and resources. Remember: the ultimate goal of machine learning is to improve investment performance. Compensation should reflect the potential improvements machine learning can bring to your business – even a small percentage increase in your returns could be worth millions of dollars.
Invest in top-tier talent to ensure your machine learning initiatives yield substantial returns: ballpark salaries for data scientists and machine learning engineers start at $150,000 to $250,000 per year, plus performance-based compensation.
Use a specialist recruiter
Many firms and employees claim to have machine learning in their repertoire when in reality they’ve built glorified databases. But, unless you’re a machine learning expert, chances are you won’t know how to evaluate potential hires and differentiate between the fluff and the diamonds.
Testament to how acutely this issue is felt is the fact that 39 percent of the CRE investment firms surveyed by Altus cited hiring and retaining qualified staff as their most significant challenge when building in-house machine learning capabilities.
To solve this, consider hiring through a recruitment firm that specializes in technology and machine learning talent. Again – not cheap, but if you’re going to invest in machine learning, make sure you do it properly.
Instill a culture of collaboration
Once you’ve hired a tech team, it’s critical to foster an environment of collaboration within your firm. The investment team needs enough insight into the tech team to trust the data and understand it, while the tech team – who typically won’t come from real estate backgrounds – needs to get familiar with the space they’re working in.
Final takeaways
Just 30 percent of the North American CRE firms involved in Altus’s “The State of Data Science in CRE Investing” survey reported having dedicated internal data science teams – in contrast to 65 percent of APAC firms. By leveraging machine learning now, you can create a real competitive advantage and stay ahead of the curve.
