Finding deals using free data
On this project, I worked with my brother Nick Lin. We both love data science and thought this would be a fun portfolio project to scrape public government data or websites like Zillow. For more technical findings, this is cross posted on his blog .
1. We started off poking around on the public property tax information for Chicago. If you google for your "county/city name + property tax assessor" it will likely be the first thing that comes up.
What we found was that there's a way to gather every bit of tax information on the property available essentially, such as when
- the last time they paid the tax
- the size of the building and lot
- the class the building falls under in the tax code (residential/commercial/warehouse/etc)
- the Property ID number the city gives the piece of land
From this alone, we created two maps where you can see the income disparity and property class breakdowns in Chicago. Essentially an expensive north side with limited long-term hold discounts, a deeply discounted south side. A bunch of industrial properties along the river. How you want to interpret it is up to you, but it's a map that can help narrow down the neighborhoods to purchase in.
2. The property ID is important for us to cross-reference to who the owner is, their address, and contact info which is logged under public data. When you record a deed or you transfer a mortgage, all of that gets logged under the deed recorder website of your local government. It also records deaths and divorces.
This lets us know if the owner is in a precarious position where they might want to sell their property quickly in exchange for larger discounts. This lets you avoid the agent's fees, which often will take up 6% of the sale price, which is already a significant reduction in price that the seller can pass on to you.
This includes
- people who've paid off most of their mortgage
- people hit by very large property tax increases
- people going through a divorce
- owners who live out of state
- etc
The best way to hide it is to get an LLC and have that be the owner on the deed records. Then use a number from a free google voice account, that way you can abandon it if need be. However, wouldn't you know it, some states will publish LLC information too (cough cough California). There are ways to be anonymous that I post on my blog and a prior post.
Services like Reonomy exist, but we didn't want to be limited by data format for 500$/month while working with a subpar algorithm that everyone else also has access to. The goal is to start cold calling and eventually building our own prediction algorithm based on success rate.