Using R to Model UK Residential Property by Giles Heywood

[This article was first published on RBlog – Mango Solutions, and kindly contributed to R-bloggers]. (You can report issue about the content on this page here)
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.

Our first speaker at London R is Giles Heywood who works as Chief Data Scientist at Seven Dials Fund Management. As an alternative property specialist, he uses model-driven strategies to support residential property investment – and as a user of R for 20 years, and author of ‘its’ package for irregular time-series (published on CRAN), he naturally turns to R for all analysis.  

Once a proof of concept, his robust and optimised product readily models district, area and regional property trends, cycles and risks. 

How can the right data support property choice?  

There is a growing appetite among investors for real estate alternatives – including student accommodation, senior housing, build-to-rent residential and hotels, which can offer better prospects for income growth. It also offers risk-adjusted returns than the traditional commercial real estate segments.  

The 7-strong team at Seven Dials Fund management takes a structured and systematic approach to direct real estate investment and also indirect investment through funds in this way.   

Whilst commercial real estate lacks comprehensive open data on transactions, residential property benefits from transparent and complete data on crucial variables of transaction price, floor area and income data to model the dynamics of affordability.  

Our approach is a meticulous analysis of the systematic drivers of return and the regular and often predictable patterns generated in long cycles. For a first-time buyer that can choose a property between small and expensive and larger but cheaper, the right data could help the most appropriate choice and its impact on the future property ladder progression.    

Is modelling in a property-related application fairly unique?   

Although Seven Dials primarily advises institutional clients on large portfolios, some of the most exciting opportunities are in delivering quantitative insights to homebuyers and in particular high net worth investors. We see important synergies or at least significant overlaps between institutional and retail.   

For many, buying their property is one of the most significant financial decisions they will make. Imagine if data science could be used to support decision making in line with their mortgage in the future. The housing market has generally gone up since the key ‘Price Paid’ dataset appeared in 1995, however in 2008 we saw falls of 15-20% nationwide, and in some areas prices have only recently regained 2007 highs.  Both the relative returns and risks can be tracked, modelled and managed. Of course institutions have models for property risk and return, and had sophisticated models back in 2007 which to some extent failed in the crash.  Technology has moved on considerably, aided by t-copulas, non-parametric bootstrap and stress-testing. What our team has done is not to copy others, but to start from the ground up with the best repeat sales indices we can construct, factor risk models, and forecasting consistent with those foundations.   

And how are data science models used for residential property investors now?   

There are some prototypical models on the major portals, and one of the most popular is automated valuation models (AVM).  We don’t do that, for all kinds of reasons, but it’s very appealing for individuals to get updated valuations on their homes and maybe on others, including those that are not on the market.     

What will your talk focus on and what might be the key take aways from your talk? 

My specific contribution to modelling at Seven Dials is projecting relative return within sectors of residential real estate to an investment horizon, using factors. The first factor is the overall market direction and is the sort of macroeconomic variable that is quite hard to predict, so for example an unforeseen pandemic did not hold back the market – to the surprise of many. However the relative price performance is more foreseeable since it is essentially driven by microeconomic forces, and in particular by affordability.   

In addition to the models’ straightforward price-forecasting applications for homebuyers the same analytical framework will be familiar to institutional investors and lenders, and can provide strategies for risk-controlled portfolio management.    

I’ll take you on a highly focused and structured trip through a stack of three models and show how they relate both to familiar ideas like the ‘ripple effect’ but also give precise insights into a long cycle driving relative returns both locally and nationally.  Everything is in R, and I’ll link it to some of my package choices for getting both coding and analysis done fast and accurately, or at least I can answer questions about that.  

Using R to Model UK Residential Property by Giles Heywood 

Will you be joining us at LondonR ? Giles Heywood who works as Chief Data Scientist at Seven Dials Fund Management uses model-driven strategies to support residential property investment. In his talk, discuss how both the relative returns and risks in property investment can be tracked, modelled, and managed.  Join us at LondonR  

The post Using R to Model UK Residential Property by Giles Heywood appeared first on Mango Solutions.

To leave a comment for the author, please follow the link and comment on their blog: RBlog – Mango Solutions. offers daily e-mail updates about R news and tutorials about learning R and many other topics. Click here if you're looking to post or find an R/data-science job.
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.

Never miss an update!
Subscribe to R-bloggers to receive
e-mails with the latest R posts.
(You will not see this message again.)

Click here to close (This popup will not appear again)