AI:Immo

A service for searching attractive real estate objects at judicial auctions with the use of machine learning techniques. Potentially huge market: 1.5 billions € only in France for the last three years. Easily expandable business-model.

Already collected a database of court auctions’ announces and results. They are manually marked by 160 parameters and prepared for training. Few models are trained: one for predicting winning bid, others for predicting if winning bid is more or less than minimal market price.

30 000 real estate agencies in France
x 80% in just two regions (Paris metropole and Provence-Alpes)
x say 10% of whom are interested in the service
x 1k to 10k € subscription fees a year
————————————————
2.4 to 24 mln $ of sales a year
That’s about the searching service.

But the market of the real estate objects sold at courts auctions is pretty big: almost 350 mln € in 2018 in France and more than 1.5 billions € in the last three years.
Buying, say, 10% of these properties with a sufficient discount can show a decent return on investments.

So there’re at least two options for developing this idea excluding a quick exit option — fast sale of the service to an interested data aggregator.

Some real estate objects are sold with big discount to the market price. There may be found announces with a discount up to 30% to the market price. Such discount doesn’t guarantee profitability of a buy, but increases chances.
A lot of data about court auctions is accessible but isn’t entirely used by market participants since they are mostly small family-held companies, which can be a huge competitive advantage of the start up.

The solution is to create a database and a model for predicting winning bids and for finding the most attractive real estate investments. Such service may be interesting for real estate market participants like real estate agencies, brokers (over 30,000 companies in France), for court auction announces aggregators and publishers or can be used by the start-up for its’ internal purposes like investing in such objects by itself

Mostly small family held building companies and to some degree real estate agencies

Since it’s a market dominated by small companies facing capital and liquidity constraints, they are forced to invest in a small amount of buys every year struggling to make every transaction profitable and heavily investing in information asymmetry which makes them even more concentrated only on few deals annually and limits even farther their potential growth opportunities depriving them of the benefits of the diversification of the economy of scale.
A company facing no such capital and liquidity constraints using cutting edge science and state-of-art technology can get over the market and eliminate the very few remaining competitive advantages of the small family held companies

Hard to predict the exact future price of the service accepted by sufficient number of the market participants, so the subscription fee from 1k to 10k a year sounds like fair enough.
The numbers of the potentially interested companies counts only real estate agencies (30 000) and doesn’t include small family held businesses buying properties at auction with the purpose of the following resale maybe after a small reparation.
We are mostly interested in those companies that are situated in two regions only because 80% of the property sold at court auction is in these two regions.
I consider my estimate of 2.4-24 mln euros of sales as rather conservative. Expansion on the entire European Union would increase sales up to over 100 mln. euros

As to the proprietary investments in the judicial court sold real estate, this would require massive financial resources.
As I’ve already written the amount of the real estate properties sold only in 2018 is about 350 mln. euros. According to my estimates based on the data collected, 65% of all properties sold are sold with a discount to the minimal market prices (according to the data about french real estate market prices I collected). Some of these objects are discounted due to their quality.
Let’s say 20% of the property sold are underestimated by market participants — so the property’s worth of 70 mln euros. The company may use financial leverage and partially finance buying these properties with bank credits facility, let’s say up to 40%. Let’s also suppose that the average discount is about 10%.
So buying 70 mln worth properties we’ll resell them for a price of 10% higher.
It could give a return on equity of 15% a year.

There are two option:
Either subscription fee based search service
Or proprietary technology based investment firm

1/ developing instruments automatically downloading announces and relevant market price data
2/ developing instruments for preparing announces to training
3/ developing search and visualization instruments
It’s back-end on which can easily be based the research service as well as proprietary software for in case of the investment
4/ perfectionning the models and the data bases
5/ hiring either employee or consultant - data scientist or data engineer.
6/ creating a team of coders
7/ registering a company in Luxembourg
8/ hiring a CFO with background in startups or real estate investments in case if we decide to develop an investment company, not a service for the third parties
Depending on the sum invested and decisions about future of the startup made:
9/ hiring french real estate specialist including technical specialists and scouts
10/ starting collecting similar data base for the United Kingdom and Germany

Say 200k for 15% payed in few transactions. Monies are going to be spend on the first seven point written in the previous section. I’m 100% pro an anti-delution clause (to a degree, of course) for the very first investor.

Huge global real estate data aggregator like Zillow may decide to come in to the market. Zillow already tests so called one buy click allowing its’ users to sell their property to the company literally in one click. It’s question of time when they realize the perspectives of the market.

Not invention or patent per se but the data base is protected by the data base developers rights