matix.io

We built an application that helped real-estate investors find leads on pre-foreclosure properties, replacing a task that could cost $900 per month with a simple tool.

No one wants to spend their days reading through 10+ page court documents to a few addresses and dates.

Trust me, I did it. 

I hand-picked over 1,000 pieces of data to teach a machine learning algorithm how to do it on its own. It was painful. It takes a long time to download a PDF and scroll through the pages looking for a small piece of information. It's difficult enough to do once. Repeating the task for hours is mind-numbing.

What is pre-foreclosure investing?

In the United States, when a borrower misses their mortgage payments the home goes up for public auction, known as a Trustee's Sale. Before the Trustee's Sale, the borrower is served two documents: a Notice of Default and a Notice of Trustee's Sale. The Notice of Default is a warning for the borrower, and is normally served after 90 days of missed payments. The Notice of Trustee's Sale sets the date, time & location of the Trustee's Sale. 

Pre-foreclosure investors will look for homes in the foreclosure process and approach the borrowers with offers before the home makes it to public auction at the Trustee's Sale. The goal is for the borrower to receive an offer that will help them out of their position, while the pre-foreclosure investor receives a deal on a property.

The pain of pre-foreclosure investing

The Notice of Default and Notice of Trustee's Sale are often publicly available at the County Recorder's office. The system in place depends on the county the home is in.

If the documents are available, they are often scanned PDF documents. Investors will spend hours downloading the 10+ page documents & looking through them for valuable information. And unfortunately, that information isn't easy to find.

Hours of time, saved.

I wanted to save some time for real-estate investors. Here's how I saved time:

  1. I built a robot that would automatically retrieve every document from the counties records. Already this would save about one minute per document. In busy counties, that could be 30-60 minutes per day.
  2. I used OCR and machine learning to take the data from the documents and put it into a database. This would save around 5 minutes per document. That's 2.5 hours - 5 hours per day, or a part-time employee!
  3. I made a search tool, to help investors find exactly the information they were looking for.

Paying a part-time employee to search through these databases would cost a minimum of $225 per week, or $900 per month. That's a lot of value!