Big data is one of the key buzzwords in all industries. Especially analysts in the financial industry have to cope with large amounts of data to conduct their research. One of the main struggles of all data scientists and analysts is the fact that the quality of the raw data is, in most cases, not good enough to conduct meaningful research. This means that they need to do a lot of work on sorting and cleaning the data so that an algorithm, or another kind of artificial intelligence (AI), can finally conduct a meaningful analysis. This is also true when it comes to the analysis of mutual funds or fund markets in general.
What Does This Mean for Quantitative Fund Analysis?
Even as the analysis of mutual funds may look quite straightforward, since mutual funds are regulated investment products that have to meet regulatory standards regarding disclosure of data, one has to structure the respective fund data to conduct meaningful reports on fund markets or to build fund recommendation lists. The first step is to group the funds by their investment objective, to build meaningful peer groups.
This may sound somewhat easy, as this means at the top level that the funds get grouped by their asset type. With the next step, the funds have to be grouped by their investment objective, such as equity U.S. versus equity Europe, to enable an apples-to-apples comparison. But even this second level would not lead to meaningful comparison, as the investment universe of the funds often varies. With regard to this, the next step would be to separate small apples from large ones, such as equity U.S. small caps versus equity U.S. large caps.
That said, even these steps will not lead to sufficient results since mutual funds do often follow a specific investment approach or strategy—such as value, growth, dividend/income, etc.—which determines the risk-return profile of the respective fund. The implementation of further strategies like ESG/SRI or any other sustainable investing approach makes this analysis even harder, as all of the different approaches lead to specific risk-return profiles and/or foster specific investor preferences. Therefore, this last step is the hardest to do, as this would require a look through the holdings of any fund and an up-to-date data set with company valuations and ratios to assign the single securities in the portfolio to a specific investment style and/or strategy.
As these are, in general, the steps for equity funds, it is clear that this becomes even more complicated for bond, multi-asset, or alternatives funds, as these funds can use multiple investment strategies and derivatives in one portfolio.
Even as these basic steps are already challenging, the global investment industry is permanently evolving and launching new products with new investment objectives or strategies, which means that fund analysts or their data provider have to update their fund classification systems on a regular basis.
What is Needed to Run a Meaningful Fund Market Analysis?
If one wants to conduct an analysis of different fund markets, it becomes even harder to structure the data in the right way as, for example, the domicile of a fund might be different from the domicile of the fund promoter or fund management company. A study based on fund domiciles may, therefore, lead to incorrect results when it comes to market sizes, distribution opportunities, or competitor analysis. This means the respective analysts have to conduct proprietary research and structure the data manually to create meaningful data sets that highlight the right spots in the respective research reports. In addition to this, it is obvious that the respective calculation methodology for fund flows and the quality of the underlying data are key contributors to the value of this kind of analysis.
The views expressed are the views of the author, not necessarily those of Refinitiv.