October 24, 2019

2 min read

Fintech and Innovation

Quant investing 101: Clustering
We’re back with another article of our Quant 101 series.
Nicholas Flaherty Image

Nicholas Flaherty, Investment Strategist at FWU Invest S.A.

Quant investing sounds complicated, but it doesn’t have to be! In this short series, we will delve into some of the key processes behind our Quant methodology and hopefully enlighten you and show that this is not a black box! 

One of the first techniques we use when constructing a portfolio is called ‘clustering’ – it is essentially the first step. 
So, what is it exactly? 
The best way to think about clustering is that it is a novel way of dividing up the stock market universe. Traditionally, the way investors have divided up their investment universe is by putting stocks into regions or sectors: classifying a stock whether it’s based in Europe or America or whether it’s a consumer company or an industrial company. 
This way is not bad, but is not exact enough because investors often end up putting companies in the same basket with others that do not belong together. For example, Amazon is put in what it is called the Consumer Discretionary sector – but here it is compared with McDonalds and GM. But does it make sense to place one of the most innovative and multifaceted companies in the same ‘basket’ with these other two retailers? In our view, no; we would much rather place the likes of Amazon with companies with which is it is more readily comparable. This, then, is the essence of clustering: ensuring that stocks are placed in more exact baskets, so apples can be compared to apples and not oranges.

The result of this is a ‘cleaner’ way to look at the stock market universe. We reach a scenario where we can compare apples with apples within clusters but at the same time can create a portfolio made up of baskets of stocks that are the most uncorrelated with each other. By reducing overall correlation in the portfolio, we improve diversification and reduce risk. 
Clustering, then, is the foundation of our Quant methodology, whereby we place stocks into the baskets they belong and hence ensure basic quality to the portfolio.