Report "Using AI to Optimize Rental Investing"

Severin Sadjina
Image via

The speaker was Dr. Severin Sadjina, a long-term rental investor from Norway. Severin shared how he leveraged a machine learning to optimize his REI journey.

Table of Contents

Video Report


Left click to see the transcript

Dr. Severin Sadjina:

Three years ago I was getting ready to pull the trigger on my first buy and hold real estate deal. I was about to make my offer but I felt really nervous. On one hand, I was nervous because I was about to send off all the money I had in the world back then. Was everything I had saved up over seven years prior on a very meager salary from my academic positions as a physicist.

And on the other hand, I was even more nervous because I started having doubts about my approach. What if my AI models were wrong? What if this was not an undervalued deal, but I was in fact overpaying? Or what if the model's estimate for the rent was completely off and the deal would just bleed me money?

You see I have a small confession to make. In the beginning of my real estate journey I had no idea about real estate. None. I didn't know the markets and which ones were best. I didn't even know what questions to ask to find out. I did know where I wanted to go.

I wanted to build a portfolio of rental properties to achieve financial freedom. But I had no idea what strategies would get me there reliably. And I also didn't know how I would find good deals easy and fast.

I did what any theoretical physicist would do in my place which is build advanced models and look for patterns in the data. If you give me eight minutes now, I will show you how artificial intelligence and computer simulations can help you to find the best deals easy and fast to optimize your strategies to maximize profits potential and control risk, when building a portfolio and how to gain interesting and valuable market insights. And because AI and simulation and real estate are all topics with at least a billion facets, each I will stick to my personal journey here.

I will show you how I use these technologies for my purpose to simply wet your appetite and to try to prime us all for an interesting and exciting exchange later.

To help me find deals I use two types of machine learning models. One type gives me market value estimates for any type for any property. I know what kind of equity or discount I can expect when I buy something. And the other type of model predicts rental income for any property. I know if the cash flow is good enough. And together with the data I collect from property listings . This gives me all the information I need ready in one place to know if something is a deal or not. For example, I see cap rates and how much money I can expect to make going in. More so it enables me to automatically generate deal reports such as the one you see on the right, which I can then immediately send to the bank to ask for financing or to partners and investors.

Early on in my real estate investing journey I was also faced with the question of what portfolio to build? And how I can best build it? Real estate investing is not rocket science. It's mostly simple algebra really. But you quickly get the questions that are far from trivial to answer. And here are just a few of the questions I encountered early on. What loan type should you get?

Let's say okay. Let me back in a little bit. Let's say you want to build a rental portfolio of, I don't know, 20-50 single family homes or something:

  • What loan type should you get on each property?
  • How much leverage should you use?
  • What cap rates do you need?
  • How much discount do you require on each deal?
  • Is it better to sell the properties after a while or should you hold on and instead refinance to buy more?
  • If you hold how long should you hold for?
  • Towards the end when you want more cash flow should you, for example, sell two thirds of your portfolio to pay down the depth of the remaining third?
  • How much capital do you need each year?
  • Should you team up with investors or do it all by yourself?
  • What happens if interest rates go up or down for that matter?
  • What if inflation changes?
  • And of course, what if the market goes down?

You quickly get into territory, it's really confusing to know how you best build your portfolio to best optimize profit potential and to best control risks. At some point I just had it not enough. I sat down and I simply wrote my own real estate investing simulator. And with it, I can simply try out a million scenarios literally and just simply see what works best under which conditions. And what this gives me is clarity for what is most important during any given situation, knowledge of how I can maximize profits, and just the right controls to keep risk at a minimum.

A scientific approach like this using AI and simulation paired with the right data can also reveal a lot of interesting and valuable market insights. One obvious example is to simply be looking for the best cap rates in the country. (In my case here this is Norway). And luckily it turned out that my, uh my local market that's the green. No, it's close to the green dot anyway. But it turns out that my local market actually happened to be quite good in terms of cap rates, which was exactly what I needed for my strategy. That was I guess just pure luck. Or how about knowing exactly what month of the year, most people are looking for a new place to rent in the neighborhood for any specific type of property.

This graph here shows you how long it takes for the average rental ad to be snatched from the market and taken offline again at any month of the year. Here, for example August and September seem to be best. Or how about monitoring how much living area in total there is to buy on your market over time? This allows you to discover trends and to potentially gain an edge over the competition. This is an example here on the screen of my look for a certain type of property in my local market and how much living area in total was out on the market to buy.

You can see some seasonality. Once a year you have these dips and these highs. But you can also start to discern maybe a slight downward trend. If on the other hand, this is going up. And there's more and more living space available for purchase. Then you might start to be careful and move somewhere else maybe. I mean not moving somewhere else but move your investing somewhere else, I meant.

So, no More Humans? (Titles)

Okay, so technologies like AI and simulation give all the answers and make humans obsolete. Well no. I still double check my models. They can miss obvious things sometimes. For example, it's hard to include a property's condition. Always has been continued to be. It's also very rare, generally. <illegible> It's also very rare to deploy machine learning models and not having to update and maintain them from time to time. It's an ongoing process, it's ongoing work and of course you need to know what you're doing. And maybe most importantly, real estate is a people's game. And I don't think that, uh that will change anytime soon and also think that we wanted to change.

99.7% ROI (Titles)

A little while after had bought that first property I was on my way to see a client. I was outside walking. This was in January in Norway, it of course was freezing cold. But I felt very warm and fuzzy inside because I had just checked my bank account. I had all the money put into that very first deal. I had just gotten back out again from refinancing the original loan. Of course, it's always nice to get some money. But what this really meant was that my models had been right. That it had worked and it had worked exactly like I had planned.

What Now? (Titles)

It's all right so you're probably feeling a little like this right now, after all this information super high speed. But I do hope that you also feel curious, excited and maybe a bit inspired. And I hope that you now understand that new technologies like machine learning and computer simulations they do hold enormous potential to find great deals fast and easy, to optimize strategies and portfolios, and to gain valuable market insights. And now I can't wait for an interesting and engaging exchange on this topic with you all. Thanks!


Kevin Hofstee: Thanks everyone for sharing. That was great so do you. I was curious how long you've been doing this? And I like how many just like talk a little bit about yourself? How many deals you've done, etc.? Just more curious.

Severin Sadjina: Yeah, I would still say, I'm a total noob. I started as I said I bought my first deal a little about three years ago. Now I started using Machine Learning to study the market around, yeah, pretty much four years ago now, a little over four years ago now. And I have so far bought four deals. Now I have a portfolio for rentals. There's still a lot of space upwards.

See the rest of the questions and answers in the video beginning at 10:13.



Would you like to receive articles like this? Subscribe to our mailing list right now.

You may also like:

Château de Chaumont-sur-Loire

700 naming schemes for rental properties

Real Estate Investors can give their properties short, catchy nicknames. There are about 700 naming schemes. The article compares different naming schemes.

Stairway to heaven in the ancient city of Zefat

Rental Ladder Rounding

Real estate investors often round up rents to the nearest hundred for analysis. Such rounding can lead to catastrophic errors in real estate investment analysis.