Chapter 1 : Index construction
Companies are part of related standard industry classifications or produce a keyword hit from an NLP process but only 58 make it to the index
an entire family of thematic indices . For example , 415 of those 466 companies are included in the most granular of standard industry classifications related to digital advertising , but only 10 % of those are truly pure plays .
If we relied only on those classifications , we would end up including companies like Omnicom Group , which is a fullservice advertising and public relations agency . This would be an inappropriate exposure that an investor would not choose as a way to express a bullish view of the digital advertising space . Omnicom Group is roughly flat over the last five years while the BlueStar Digital Advertising index is up approximately 70 %.
On the other hand , 15 % of pure-play digital advertising companies are not included in any related standard industry classification and would not be considered without incorporating an NLP process . At the same time , 381 of those 466 companies have keyword hits from an NLP process but only 2 % of those are pure-plays in digital advertising .
Although companies may describe themselves as being related to digital advertising or being heavily involved in it , they might not be a fit for the index . In some instances , even company-reported information is difficult for a researcher to interpret , let alone any sort of automated machine-learning programme .
Typically , this is the case when a company does not report any segment-revenue breakdowns or the reporting segments are too broad to draw direct conclusions from . For example , a company may report that 75 % of its revenue is from its “ services ” segment . Here we would need to refer to the company ’ s description of customers and services offered , as well as customer case studies to determine what portion of its services segment is related to digital advertising .
Thematic indices should allow any type of investor to act on their analysis with precision . But the returns of an investment will only correlate with the thesis if the index used is pure and complete ; a properly designed thematic index requires a research process to ensure the attributes are achieved .
So the next time you are hit by lightning and come across the next big investment idea , do not leave money on the table or anything to chance . Sure , you can do some initial research and buy a couple of individual stocks that feel like a fit . But the more sophisticated and highconviction approach is to use a well-constructed thematic index .
Joshua Kaplan is global head of research and investment strategy at MarketVector Indexes