An Engineer’s Hack for Predictable Real Estate Performance

When I started my real estate investment business in 2005, my engineering background led me to rely on analytics to find high-performing properties. After months of testing, I realized something important: people don’t choose homes based on logic; they choose homes based on emotion. So, traditional analytics will not work.

Then I (finally) decided to focus on historical tenant behaviors. In other words, I stopped looking for why properties perform (since I couldn’t figure out a reliable pattern) and started looking at what performed. Let me explain with an example.

Suppose there are two countries: Country A and Country B. Your task is to determine which of the two countries is more desirable to the people of both countries.

The Analytical Approach

Suppose you have every piece of data you want about both countries, unlimited computing power, and all the data scientists you want. What are the odds that you could accurately select which country the people of Country A and Country B prefer to live in?

In my opinion, about 50/50. Why? Because you do not know what the people of Country A or Country B value. It is very hard to “quantify” emotions. The failure of traditional analysis to accurately predict property preferences is why my initial research failed.

The Behavioral Approach

What if you counted the number of people moving between Country A and Country B? And, if the people moving from Country A to Country B is 10 x the number of people moving from Country B to Country A? You know Country B is preferred without needing to determine the basis of their decision. Observation is often the most reliable method if you are dealing with non-numeric decisions. Another advantage of observations is that the behavior of a population changes very slowly.

An Engineer’s Hack for Predictable Real Estate Performance

I started studying our target tenant segment (that itself is a topic of another blog) using historical rental data from the MLS: which subdivisions have frequent tenant turnovers (frequent rental re-listings), which subdivisions have long days to rent, which have higher $/SF, etc. etc. The two most important factors for investors are time to rent and length of stay.

My research into behaviors continues to this day. For example, in 2008 and 2009 (during the height of the Financial Crisis), I studied which properties rented and which ones took forever to rent. This required a lot of driving around and trying to figure out what the difference was between properties. While I didn’t always understand why the segment behaved the way it did, the correlation between property characteristics and tenant preferences was consistently strong.

Some examples:

• The lot must be bigger than 3000 ft².
• It must have a two-car garage where the cars are parked parallel. There are properties with two-car garages, but they park nose-to-tail. These do not rent well.
• Properties with all electric appliances take up about a month longer to rent and rent for less than properties with gas appliances.
• The size of the master bedroom matters. If any dimension is less than 12 feet, it takes longer to rent. For example, suppose you have two identical properties, and the only difference is the dimensions of the master bedroom. Suppose one is 11’ by 18’ and the other is 12’ by 12’. The property with the 11’ x 18’ takes significantly longer to rent.
• We’ve identified over 2,600 streets where properties do not rent well. On some of the streets, I can understand the reason but most I do not.
• Some subdivisions have average tenant stays significantly below our 5-year average. Many of these subdivisions look the same as the ones that perform, but the difference in the length of tenant sty is significant.
• There are many floor plans that do not rent well. We know of many and eliminate them from consideration.

So I focused on identifying the indicators of high-performing properties and stopped worrying about the reasons behind them. As an engineer, my priority is building a system that works reliably, not explaining why it works.

Today, we have about 40 criteria that each candidate property must meet to be considered.

However, while most behaviors are almost constant, there are some that change slowly over time.

For example, until about five years ago, we never considered two-bedroom homes. When they started appearing in our Investor Tool results, I wanted to know why. What stood out was that not all two-bedroom homes rented at the same pace. After comparing many of them, I found the difference: properties with a well-defined office nook rented quickly, while those without did not.

After more research and discussions with property managers, my suspicion was confirmed: many Millennials prefer living alone, and they’re often drawn to two-bedroom homes with lower rent—especially when one area can double as a comfortable office.

Summary

I’ve analyzed countless data sets using traditional techniques I learned in engineering school. But people don’t behave logically; they make decisions based on emotions. That’s why our software consistently outperforms others: most systems rely on traditional analytics alone, while ours is built on historical tenant behavior.

I’m currently writing The Engineer’s Guide to Safe Real Estate Investing, which will be available for free soon. In it, I’ll share the engineering framework my team and I developed—and continue to refine—to consistently deliver high-performing properties with low risk and minimal hassle for our clients. I’ll let you know as soon as it’s ready and how you can access it.