Proptech — Can AI Help Property Developers and Investors Make Money?
A few years ago, after dropping off my son at school, I got talking to one of my fellow drop-off dads on the walk home. As usual, the conversation turned to questions of what it was we did for work? He was a retired mathematician…..insert a long awkward pause…. “Did you teach?” I had already ascertained that he was younger than me, so I had to know how he had achieved this remarkable feat so young and was still able to afford the rather impressive Hampstead house that we were now standing in-front off.
It turns out he had developed an algorithm for hedge fund investing, which worked so well that he and his partners had taken the incredibly large sums of money they had made and decided to retire. All those times I had sat there in maths class thinking ‘how can this ever be useful in the real world?’ Here was my answer.
In this case it wasn’t really AI, it was a mathematical algorithms that had given him the edge on his investments. There is a difference and it is only recently that AI has been used to support hedge funds and other stock and share investors. So instinctively it would seem that AI is the obvious solution of choice for taking investment algorithms that one step further.
In 1995 at university studying AI the 1st thought of every student when set their final year assignment was — wait a minute — huge amounts of data, neural networks — easy I am going to predict the stock market, make a fortune and retire some hours after receiving my degree.
It turned out not to be quite so easy. Too much data, or not enough accurate useful data, too many variables and not enough processing power meant it was as insurmountable a problem as predicting the Lottery numbers.
The story is not much different today. Voleon an AI stock predictor lost money between 2008 and 2011, and while having some success between 2011 and 2015, again lost money in 2016. While, some new AI investment technologies are emerging, it seems they are still a long way from having any success, according to Bloomberg AI hedge funds had their worst month ever in Feburary 2018.
So is there a way to use AI to help property investors, where there might be fewer variables? I interviewed Marc Herman, at Greencourt Property, a medium sized property investment company based in London, who ran me through his model for investing. It became quickly apparent just how much of what he does is based on expertise, experience and instinct. When choosing an investment strategy according to Warren Buffett, it is not just about the numbers, it’s about understanding what the business does — “Never invest in a business you cannot understand”. Just in the same way for Marc, it is about understanding the property, the restrictions, its viability, the people involved, local knowledge and the current state of the market.
These are not criteria which can be easily adapted to current Artificial Intelligence techniques. AI is great at finding patterns in huge amounts of data but what it is not good at are those human traits such as instinct and experience, thinking differently or latterly as well as making connections between obscure pieces of data that require digging a bit further than the data with which you are presented.
This is the fundamental difference between AI and real human intelligence — there is no out-of-the-box — it’s all in the box, hidden in a mass of complex equations. I am sure we could account for some of the factors that are not purely mathematical but making deals is often an emotional decision making process and we have yet to come up with algorithms that account for these variables.
However, there are areas where AI can help. One possibility could be to address the limitations on how many properties an investor can actually consider and carry out some level of due diligence on. The inbox fills with new properties and the agents call in with their latest offers, the volume becomes daunting and many will not even get the briefest of reviews. The investor becomes less able to find and categorise those properties that might be of actual interest. This leads to them missing out on deals that might just turn out to be perfect for their portfolio.
Designing an AI system to scan properties on the market based on a set of investor specific criteria might be a useful addition. An AI filter could take into consideration factors such as potential incomes and yields, environmental considerations, such as other local businesses and properties, demographics, transports, schools, whether the area is up-and-coming and the costs to develop. Acting as a gatekeeper and assistant it could quickly weed out those that do not meet the investors criteria and create a list of those that might just be worth the effort of further research. It may even be possible to use this to find properties that are not even on the market yet but would be of interest should they become available.
For the small investor and for those companies with huge portfolios and vast sums of capital that need to be invested it could be a very useful tool.
When it comes to property investing and developing, human factors are still strongly at play, understanding the seller, knowing when to sell or develop, the tiresome and overly manual planning process.
The deal is still the deal and AI is long way from the human skills required to negotiate the deal, it’s that X factor that every good dealer appears to have.
So my simple algorithm for a good property investment strategy is:
X + AI/f = P
Where X is the Human ‘X’ Factor, f is a factor by which AI can assist the Human and P = Profit. No Nobel Prize winner and I probably won’t retire on this one but if you are a good
investor, even in a world of robots and AI, you still hold the key to turning a profit.
