Utilizing Data Analytics in Real Estate Investing: My Journey from Failure to Success

In 2004, while living in NYC, I decided to embark on a new career path. To identify potential business opportunities, I considered my most frustrating experiences, quickly narrowing the focus down to investing in properties.

Unfortunately, all real estate agents could do for me was send me MLS data sheets for properties I selected; no analytics, processes or services could be offered. Therefore, I had to manage everything myself which took up considerable time and money in terms of correcting mistakes later.

So I see an opportunity in business; now, however, I need to know where best to set it up (other than New York or New Jersey).

How I Conducted My Analysis
For my analysis, I started researching how retail store chains choose sites for new stores. Based on this research, I determined a sequence of events needed for their success as shown in the chart below.

My initial decision was location; after conducting extensive research, I selected Las Vegas.

Step two was to identify an ideal tenant pool segment to target, as this step is essential in order to secure reliable income from properties. A reliable tenant typically stays for multiple years, pays their rent on time every month and takes great care in looking after it.

Based on my experience and what I’ve seen from others, reliable tenants are far and between. Since my clients and I plan to hold these properties for an extended period, we will require multiple reliable tenants throughout. One way of increasing our chances of always finding reliable tenants is purchasing properties that attract people from a segment with high concentrations of trustworthy individuals.

My task was to identify an attractive tenant segment with a high concentration of reliable tenants.

As an engineer, I rely on traditional data analysis methods. After exploring various data sets – both paid and free – and writing software programs to analyze them, however I eventually concluded that classic data analysis wasn’t going to cut it as humans don’t behave algorithmically and so this approach no longer sufficed. So I took another route instead.

Next, I set out to analyze historical rental data in order to gain an understanding of past tenant behaviors. I downloaded around 10 years of MLS rental history data and began over again; trying many things that ultimately failed before plotting monthly rent versus length of stay as my indicator variable.

I found an effect similar to what was displayed in the chart, showing a strong relationship between length of tenant occupancy and rent levels. This was exactly the beginning point I had been searching for.