In this episode of Cash Flow Pro, we talk with Stefan Tsvetkov of RealtyQuant. Stefan was a financial engineer for ten years, jointly managing a $90 billion derivatives portfolio with colleagues. Today, he is a multifamily investor, analytics...
In this episode of Cash Flow Pro, we talk with Stefan Tsvetkov of RealtyQuant. Stefan was a financial engineer for ten years, jointly managing a $90 billion derivatives portfolio with colleagues. Today, he is a multifamily investor, analytics speaker, and live webinar host to share what he knows about market data in real estate and more!
RealtyQuant brings data-driven and quantitative techniques to the real estate industry and its clients. They are on a mission to value the real estate industry through education, investment, technology, and analytics.
In this episode, we discuss:
If you are interested in learning how to use data to predict real estate markets, tune in to this episode to learn more!
Find your flow,
Resources mentioned in this podcast:
Casey Brown 0:07
Hey there and welcome to today's episode of cash flow pro your daily real estate investing podcast and YouTube channel. I'm here today with Stefan Tsvetkov Realty quant I had to I had to look down and get the name of the company because I was so focused on you not messing up the last name that I wanted to make sure that we got it right. So, Stefan, how are you today?
Unknown Speaker 0:32
Doing great. How are you Casey?
Casey Brown 0:34
Oh, doing wonderful doing wonderful. So Stephen is here, he's going to talk to us a little bit. So I think there is this little, this little bridge that's that's constantly being built between the old school ways of underwriting and community analysis. And over our overall market analysis, specific market analysis. And, and Stefan has some tools that give us the ability to do some of that digitally. And and using a number of different data points and some other stuff. And then we were talking a little bit about some of the AI that's involved on the property side. So anyway, Stephen, I'm gonna let you kind of tell us a little bit about what your background is, where you come from, and how you got started in this. And then we'll kind of go forward and let you tell the listeners, you know, what kind of stuff maybe your company offers, and go from there. So take it and take it away?
Unknown Speaker 1:38
Yeah, absolutely. So, myself, I am a financial engineer in my previous career. So I had a career in finance for about a decade. I'm Eastern European originally. So I came to the States at 22. I did like grad school in New York City, and financial engineering. And then I was working like as a derivatives trader, derivatives portfolio manager in finance, you know, for some of your audience. They may know what it is, it's kind of like trading options and futures and you know, like different financial instruments. Right. So that was my job before, for about a decade. And in the reason like, few years, I've been a real estate investor. In the New York City area, I mean, like different residential projects more as well as more recently, like commercial out of state towards the Midwest, actually. So that has been like my background as an investor. I also have a, an analytics company called robotic ones, where we do like different data driven, I do like different data driven methods, essentially. And the one product we now have for the public is kind of market data. So like, gauging, like different, like over undervalued markets, like price forecasting, and things like that.
Casey Brown 2:55
Okay. And so, what, I guess the big thing for me, and this is something that you and I were talking about. What's data aggregators are something that really, it makes so much perfect sense to me. But at the same time, I always get confused as to which data is important. And especially when it comes to commercial real estate like, like, what, give us an example or two of what data points can be used in specific markets. And what that gauge is that's useful to commercial real estate. Folks,
Unknown Speaker 3:41
that's a great question. Well, one thing on the property side that I found very useful, which is how I do my off market, commercial, multifamily, you know, prospecting, if you will, yeah. Using rental listings data to kind of gauge value add across different, you know, commercial off market commercial multifamily. Okay. So, the way that approach works, is, if you have something like data from be like costars, apartments.com or rather apartment finder, or realtor.com, whichever rental listings information you have out there, you can sort of use that data at scale across 1000s and 1000s, of commercial multifamily. And they had a did a web lecture for my own webinar where they showed, like, kind of studying 32,000 commercial multifamily properties in different markets and kind of trying to predict which ones of them are going to have the biggest value. So that's an example of a data driven coId like you said, like new approach versus like the older school can older school approach, right? The some ways in which this can be good in my experience, is well, one thing is okay, you can discover properties that have rents below beyond their immediate neighborhood and property class, let's say your rents that are that are out there. So they're sort of for the same property costs, same neighborhood, you know, having rents your market is the more obvious component. But then there's more insights that one can derive from rental listings information. And this kind of a unique approach, actually, I mean, I don't say unique, but a relatively, it's not something I have seen out there as far as an approach to studying commercial multifamily, I feel. But it's, but I think it's, it's a good one. And so in so another, another thing is, like, you can look into, like, what fees those buildings are charging. So like, you can look at like, are they charging pet fees, administrative fees, like all those other income components, and you can look at like, what utilities they're charging, and sort of, perhaps even from rental listings, gauge, the approximate occupancy in a way, like, based on what how many units are coming out on the market, and, and things like that. In those terms, you can kind of build a simple model in which you try to assess, okay, between all those factors, what's going to be the percentage improvement in NY that you can realize it's going to given the commercial multifamily property compared to other ones in, in the in the same market? Let's say? Yeah, yeah. And, yeah, so that, so it's like, it's a fairly, you know, not a fairly not complex approach, but it gives you in a way, a very, it's very approximate, since, I mean, obviously, you don't have the actual income expense sheet of that property. Yeah. But it's a way to sort of get a preliminary preliminary glimpse into the income expense sheets of, you know, 1000s and 1000s. Of properties. And then you can say, okay, perhaps I can, if you have direct owner components, and, you know, or agent relationships, whichever approach it would be, but you could kind of gear your marketing or efforts, you know, towards specific buildings that are, you know, bigger value in or perhaps you could also, you know, prospect in five different markets, just the top 20% in those markets, you know, yeah, yeah. Makes for Yeah, makes for the kind of more intelligent margin. So that's on the property side, like when kind of modeling of the sort of modeling of the off market inventory, rather than just like getting the data feeds and just like blasting campaigns to those data feeds, kind of like having like, more financial modeling of it. So that's one thing. As far as the market side, it's quite interesting, because, well, I mean, there is obviously some market data from costar and so forth. But now is it big enough to be able to get get like big
Unknown Speaker 7:40
information, like across the US roads, every single county, every single market and so forth, which is generally my approach on the market? I kind of I like to have the big picture for every county in the country, you know, something like this, and kind of like compare? Yeah, and so yeah, so the, what I do myself for the market side is it's, it's actually just starting the housing market, it's not the commercial multifamily inventory. Now. I also did a study of like, okay, what has been the correlation of, you know, like costars in a Rei T index, I believe, is the name and let's say FHFA home prices, and okay, over the long run, the price correlation is like 91%. So, you know, kind of over the long run, they are supposed to move together. Okay. Now, as we know, in the short term, this is different it deviates. We have like the commercial, some commercial multifamily declines after world financial crisis, for example, perhaps being more moderate than in, you know, regular home prices, terms. But again, it's just too for me to have like a relative measure and understanding across like different markets, I just look at the regular housing market. So the regular like residential properties, and the way to do this is like doing fundamental analysis on income population housing supply. Okay. And, yeah, and so I did a back study for the global financial crisis. And in then this has been the basis of like market data that we publish it. My company wrote the quant which show sort of gives, like, downside risk and appreciation, predictors for like, basically every county in the country. And so, so the to gauge downside, right, so before the global financial crisis, if one did like an analysis of income population housing supply, or let's say proxy to the more simple let's see affordability deviations versus a moving average between like, things like that. So one would, would have predicted the declines at the state level the clients to around 85% correlation, just pretty high. And one can see like actually very easily, like a like down like, many lectures on this topic, actually, at this point, where it's like, okay, well, if you look at the map of how different US states were valued at the time, Yeah, and for example, like California, Florida, Nevada and Arizona were like, the ones that decline the water after the financial crisis. And so they were valued at like, 40 to 60%, like, deviation from where they should be historically. Yeah. And like, you know, they're the clients were actually which were like, I think like 41 to 56 were the exact numbers in like those four states. And so it's like pretty in line. And then you go to like all the underwater states, which was actually even Texas was underwater with at the time, and they had like, only a 4%. Decline. And so which is like, so it was very interesting, like to see like this kind of study and, and then I did like the same study for like, every county, every county at that time, as well. And correlation drops, it's harder to predict small geographies right here, like now, like every single county correlation is now maybe around 75%. So it gets like, less predictive, but it's still, you know, you're predicting small geographies. It's like sounds 5% correlations, you know, pretty good.
Casey Brown 11:01
Yeah, seeing that. It doesn't take long to get over my head. I know, as far as when you start talking about when you start talking about a lot of what correlates with what, and that's what we're, I get lost really quick. Because, like I said, you know, before the show is like, you know, I've always wondered how some of these places use like, how many jars of peanut butter they're selling, and then all of a sudden, they can tell you, you know, other things about the market that you just never even dreamed from that one statistics, but But so, one of the biggest things that I have been ultimately, like very curious about, especially in the last like, say, probably a year to 18 months, has been how the flood of people from California has affected? Like, how do you? How hard is it to predict what communities are going to grow? Because ultimately, at the end of the day, when California start shrinking, everywhere else starts growing, it happens every, what, 12 years, 15 years, something like that. And so when those when when they start leaving, how hard is it to predict what then is going to start growing? Because, like, it's almost like a lot of those people come to tonight, I'm talking in very general terms, obviously. But a lot of those people leave there and they stop in like, let's say, Utah and Colorado, well, then the people in Utah and Colorado, go to Kansas, Arkansas, and then the people from Kansas and Arkansas, go to Nashville, Atlanta, and it just, it's like a it's like a wave. And so I think the predictability of some of that stuff has been, I think, what's escaped all of us. And I think that's why some of the tertiary markets, the markets that are outside of the main Nashville's and Kansas cities, and Atlanta's you know, the even the smaller markets, the sub sub markets have grown. And the prices and the rents are basically out of control. But how do you? How does that become either predictable or unpredictable?
Unknown Speaker 13:18
Yeah, that's a great question. Well, I think it's very hard like to your point, night, as far as when the trend changes, like when there is a change if the trend reverses in some direction, and that's very, very, very hard. Now, whenever to the extent if we take Colorado or Utah, or and to the extent that there was already a price trend or population, well, I would say a price trend is actually easier to predict. And like so let's say price trend in Colorado or Utah is kind of driven by like some of the outfox from California, you know, in that direction. And so to the extent that it's already there, you know, for a few years, then it's been fairly easy in recent years to predict the growth in those same places. That kind of might sound sounds obvious, but it's actually impressive how I did like forecast price forecasting, just at the state level, just like a simple comparison. And so at the state level, if one was trying to predict, okay, where prices grew in 2018 2019, in all those different state, Colorado, Utah, everything else, right? Yeah, the the error that I had, versus the real growth was like only 1.4% when they use like, 45 years of history. So that's the while the trend is there. That's just something to know. It's not hard, so hard to predict, it's less hard to predict prices, then our intuition suggests. Now, whenever the trend changes, we have 2021 There's massive inflation kind of falls apart. Now suddenly, there is no seven 8% It's too big. And so and so that is kind of some of that now, as far as reversing in the trend, it's it's very, very hard that will be like more complex machine. learning methods, there are various things that people do to try to predict neighborhood appreciation, there was a company called lofty, lofty AI, which is more like tokenized real estate. It's a prop tech startup. And they were doing like neighborhood appreciation based on social media signals. So what they were trying to see is, for example, if in ni, X neighborhood, if the breed of dogs, for instance change, things like that, you know, okay, let's say you have like, whichever kind of dogs now have like a new kinds of dogs that are more associated with like, higher income demographic or something, right? So it's a bit more like the change of it, you know, and things like that. So, but then it's hard, because I mean, then they're what they're actually doing is they're testing like, which of those predictors actually work, which are the ones that actually predicted appreciation? And, and so it's a lot of work in that sense. So I would say like, the reversal in the trend is very hard. As far as while the trend is there, it's relatively easy in most markets. And it depends where you are, again, if you are in Alaska or something in Alaska, the negative momentum if you will, like momentum, you know, like price momentum gets measured in statistics by something called autocorrelation. A cow correlated is like this year prices, price growth versus last year price growth. Yeah. So yeah, so Alaska had negative momentum go it by this. But many most US markets like for truancy market states, and the markets within them, like markets within Florida and Texas, and given Massachusetts had very high momentum, for momentum at the state level is like close to 80%. And so
Casey Brown 16:43
that's the momentum of price increases, right?
Unknown Speaker 16:48
The momentum of price increases Exactly. price growth. Yeah, sort of price growth in the current year versus price growth in previous years, let's say, and, and so. So that's quite interesting, because that is the case, it's actually so there's trends. So it's actually forecasting. Sometimes it's not very helpful. Now more than ever, the trend changes, it gets very, very hard. Now, if we try to forecast population, like you said, I think that's actually harder. And so if we try to forecast are people going to move to like, different places? That's a thing very, very hard. And then how much of that is going to tell us something about prices? Well, population, even at the state level, you know, how much is predicted price is like 40%. on why it's not that much. It's good. It's okay. But now you have to add other things. And
Casey Brown 17:42
you have a lot of other data, the
Unknown Speaker 17:45
other stuff and then gets better. Exactly. But in there is another thing. So another study that they did is it that's a took all the fundamentals of real estate now whether or the fundamentals, perhaps income, population, housing supply, investing, let's see what else. So let's say those are like the three from the big fundamentals, in my mind. And so you will take income population housing, so I'll try to forecast each of those separately, where population going each where income is going to grow, and so forth, and then off that to forecast real estate prices. And I did that, and I found like, five times bigger error than if I just forecast the prices themselves. So that's another like things like very impactful, because what that means is if I like to syndicator Yeah, pick your market. And they're gonna say at one to let's say, Atlanta, Georgia, because has this population growth, this, you know, income growth, perhaps job growth, you know, all these variables, inherently, the reason why they pick it is because they want that market to grow, right to experience appreciation. But in doing so, in selecting that market based on those fundamentals, they are actually committing a five times bigger error, then if they just took the prices in Atlanta, Georgia, and just assume that they're going to continue to grow the same way. And so forecasting, yeah, so it's kind of interesting. So it actually makes it even easier. Well, of course, it depends now, who were in 2012 or 11. And the market cycle hasn't even begun. There's no trend whatsoever, your only way to forecast anything would be to look at the fundamentals and forecast prices of that, sure why we are already like 10 years in the trend, eight years in the trend and so forth. The best bet for people to forecast markets, at least based on what I've seen has been just to follow the trend, you know, so like Phoenix grew and Austin grew and so forth and
Casey Brown 19:41
still growing seems like inflation didn't or more gasoline on Phoenix and made and just just explode. I mean, it's just been unbelievable. So you're here about, I'm gonna say three or four years ago. So In the individual, like the individual, real estate agents sales business, and as my listeners are well aware, I always reference this because that's, I have a great deal of experience in that side of things. But there was a company that and I don't even remember the name of it right off. But it was an artificial intelligence company who specialized in, in predicting whether somebody was going to be selling their home in the next 12 months. Now, that company was ultimately acquired by REMAX, I believe. And they had a considerable amount. And supposedly, obviously, if they were acquired by a company such as REMAX, they obviously were on to something. And there was considerable amount of stuff going on behind the scenes, but what they specialized in was, so what you would do is, is you would upload your list of, they would give you like, so many uploads a month, you could upload, like, let's say, 100 new names per month, to their, to their platform for their fee that was associated with it. And as your list grew, they would analyze those contacts. And they would give, they would give each one of them a predictability score of whether or not they were going to list and sell their home in the next 12 months. And as I bought it for two or three months, just to see what, just to get an idea. And ultimately, it was like, a lot of people, no real estate agents have no patience. So I was like, well, well, this isn't working fast enough, when, in fact, 90 days to tell me anything about anybody was was reasonable. But the thing is, is that they had these data points and supposedly had data points on my contacts, that could predict whether they were going to be moving in the next 12 months based off of I'm assuming different, like, like different websites that were visiting or things they were posting online or whatever the case is. But let's dive a little bit into that. Because I think one of the one of the biggest parts of the commercial real estate business right now is, is the is the off market property that's like the Holy Grail, the off market property that somebody's the mom and pop owned 200 unit apartment complex that's not on the market, and they're wanting to retire and it's for sale. And nobody else knows it except for this one person. That's the Holy Grail. So how does, how does? I guess they're, I guess I'm just Just curious overall, how does that work? Like? How do you predict, start on an individual level? How do they predict somebody's gonna sell?
Unknown Speaker 22:52
Well, I think it's just, I mean, it's not something because it can't specialize more like on the financial analysis side. Okay. But let's say that kind of motivated sellers. Right. Okay, like motivate yourself. So, I mean, if you have to just like I think they work for signals and kind of test how predictive they were. So if you have, I mean, let's say the most obvious, perhaps signal, and usually, it's not that but more often for off market, if something was actually an expired listing, right, or something was actually listed before. Okay, so now that's an obvious thing, okay, that person actually is motivated, he wanted to sell or they she wanted to sell, but, but it's not, it didn't work out for whichever reason, and so, but it shows like, some kind of motivation. So expired listings are a big one. And from there, you know, like, there's like all those life events, right? Like, if somebody's going through a divorce, or somebody's going through, you know, like, different hardship, you know, where's the relative depth on their property versus like, the current market values, and it's just like, all those signals that make it more and more likely that somebody would be willing to sell. I think there is quite a bit of supply on that, if you think about it, like in terms of like, probably
Casey Brown 24:04
put some context behind this real quick to add to what you're saying, I fall after I cancelled my subscription, the app, I followed it, and and I followed probably the top 10. And I'm gonna say they were about 70%. For the next 12 months, the 12 months after that, they were they were about 70% of predictability wins, I guess, if you will.
Unknown Speaker 24:35
Well, that's pretty good. Yeah. Well, I guess they I'm not sure which the app was specifically but yeah, I mean, I think that's a that's a useful thing. I would think there are many tools that do that though, in a way because there's like, even like prospect now like some of the big like, off market data vendors they that's kind of what they do, right? They just kind of deal with The inventory with the owner information and then okay, what sense is to motivation? You know, the seller? But I think it's like to the discussion just like some of those signals like are there life events? Or like how is the the overall financials of the place working? And I don't think it's too complicated to be honest. Yeah, I think
Casey Brown 25:20
well, it may not be too complicated to you, but it sure is to us. It seems like we have a it's such a interesting process. But I didn't mean to get you on to something you weren't you didn't specialize in. And so one thing that I want to the one term that you said earlier that I wanted to go back to and you mentioned costar several times. And I think costar is is one of the one of the sites that is a commercial real estate, whatever any, any type of commercial real estate activity that you're doing costar is, is is a great program and and how do Do you buy like backside data from them or something? Or is it something that's available to everybody in order to go in and look and see and talk? And what does that what do you do there?
Unknown Speaker 26:10
Well, I mean, one can house subscription to their analytics and so forth. But as far as rental listings, it's not awkward. It's not ready to offer this data service. I looked into it, and there's like a few, like smaller vendors. So I use like vendor code data affinity. Yeah. And there is also a few other like, on the property analytics, like there's very small other vendors like Mellissa. And like Melissa API, and like some other ones that kind of give you like, some of the property that if you wanted to sort of construct something like do your own Prospect now, if you will, yeah, there's like ways to sign up for like, some services, but it's not directly from costar. And I would imagine would be, like, very expensive if it was available. Sure. Sure. It's kind of more like third party vendors.
Casey Brown 27:03
Yeah. Well, you know, and there's always there's like, there's apartments.com. I mean, they have and I guess I'm assuming you can do some scraping there. If you if you,
Unknown Speaker 27:13
you could, you could scrape directly the data and also, obviously, is bigger legal liability. Or you could go to different third party vendors that they source that data in whichever way which may includes scraping. So yeah, you know,
Casey Brown 27:31
we used to think the internet made the world go round and the internet makes the world go round. But the data gives it the field to go around. I mean, it's it's the data is, has became far more valuable than I think, than any of us ever even. Yes, yeah. People that were early into the data game of trying to make things work and figure it all out was we're the ones who who really had it figured out there so so well, listen, Stephen I'm gonna I want to get to a couple of questions that we ask every guest they're all the same there's no right or wrong answer. The first question is what is the best book that you have recently read or currently reading
Unknown Speaker 28:13
good question. Well, one of the best books I read another recently like the undergrowth book called when genius failed. Yeah, like on financial markets it's a book on like the 1998 I guess like crushing debt crisis and like long term capital management collapse which was like a hedge fund at the time.
Casey Brown 28:36
And it was called what what was named
Unknown Speaker 28:38
go there when when genius failed when genius
Casey Brown 28:41
failed. What is what is dream vacation that you've either taken or hoped to take?
Unknown Speaker 28:50
Good question. Yeah, I think any vacation to a warm destination at this point. And even though it's getting warmer in New York City now for sure so
Casey Brown 29:04
yeah, it's getting warmer air was getting warmer here until yesterday, and we woke up and it was like 58 or something and I was like, wow, wintertime is bad. So so
Unknown Speaker 29:14
like the Caribbean is my favorite always and it's kind of a cliche to us people but to me, it's kind of like Yeah,
Casey Brown 29:22
but there's a lot of beautiful I you know, until the last few years, I did some map study and and so on and, and there's a lot of really like you've started Turks and Caicos and Bermuda and you can go to a lot of them are those if those are necessarily considered Caribbean or not, I guess. They probably are. But either way, yeah, it's all beautiful down through there. And, yeah, and one of our favorite places. Or one of my favorite places, rather, is was Aruba. Which I think is far south Caribbean. But anyway, tell the people how they can tell the list Here's how they can reach out and get in touch with you. I know there's gonna be several questions about the AI and about some of the data stuff that you've gotten that you have access to, especially since it directly correlates with commercial real estate. So how can the listeners reach out and get in touch with you?
Unknown Speaker 30:16
Yes. So my website realistic one.com. So I think so like I mentioned, one thing they can find there is market data for every US County, so around 2700 us counties. So there's like price forecasting. And what I find personally is more important now is there is downside risk predictions. So they're sort of if you reach market peak of Mark, go over under FERS voters is every county in the in the US and it's a very useful thing to have when picking, picking your markets today. I feel so sure. Okay.
Casey Brown 30:53
Well, that's Realty quant.com. And Why don't y'all go check out, check out his website and look at what he's got to offer. And if anybody needs anything, reach out to him or reach out to his company. I know they'll be glad to help. So Stephen, thank you so much for for visiting with us today and giving us an insight into what all of this stuff that we hear so much about at least a few definitions and things that maybe the folks can use to further their real estate education. So thank you so much. Thank you.
Unknown Speaker 31:23
Transcribed by https://otter.ai
Stefan Tsvetkov is the Founder of RealtyQuant (www.realtyquant.com), a company that brings data-driven and quantitative techniques to the real estate industry. On a mission to add massive industry value through education, investment, technology, and analytics.
Financial engineer turned multifamily investor, analytics speaker, and live webinar host. He holds a Master's degree in Financial Engineering from Columbia University, and during his finance career managed ~$90 billion derivatives portfolio jointly with colleagues.
Featured on multiple Podcast and Webinar events including Elevate, Best Ever Real Estate Show, The Apartment Guys etc. Host of Finance Meets Real Estate webinar series.