A key part of the investment management process for Asset Owners is selecting Investment Managers that through their stock picking skill will persistently generate alpha.
Not an easy task. While historical investment fund performance can be used to provide proof the investment strategy works and the Investment Manager has alpha generating skill, the key for Asset Owners is selecting Investment Managers that can consistently replicate their track record.
Measuring Performance Consistency
Persistency in alpha generation in investment fund performance is not just a reflection of superior stock‐picking skill. There are a multitude of factors such as investment management style, active share, team turnover, fund size and persistent differences in fund expenses and transaction costs that determine investment manager performance outcomes.
To identify investment funds that are most likely to meet return objectives many Asset Owners have developed complex evaluation criteria and processes. Part of this process is employing teams of Analysts and Portfolio Managers whose primary role is to sort the wheat from the chaff across the large, growing universe of investment funds.
The Investment Manager selection process is augmented by the rapidly increasing amount of data that is becoming available to Asset Owners. Nowadays it seems with the internet of things like anything about investment funds that can be recorded is being recorded into datasets.
Asset Owners typically collect their own data directly from the managers in the form of questionnaires and supplement this with data services such as Morningstar, eVestment and Mercer who collect and store significant amounts of investment fund data each day.
Measurement Challenges
Increasing complexity is resulting in several challenges for Asset Owners with their investment manager selection processes.
Firstly, the investment universe of managed funds has grown at a rapid pace. There are tens of thousands of managed funds to selection from all distinctly different and all confidently purporting to be able to meet their performance objectives.
Secondly the amount of data points that are available have also grown exponentially. The challenge is not only reviewing this data but also determining which data points have relevance to the alpha generating skill of investment managers.
Combine these challenges in the reams of marketing spin that is produced each day by investment managers and the challenge for Asset Owners can be daunting.
Robots Are Here to Help
One emerging industry trend to address these challenges is a hybrid model of machine learning and human analytics. Utilizing a hybrid model is enabling Assets Owners to augment their Investment Manager selection skill with the power of machine learning and gain an edge in a highly competitive market.
Machine learning model’s strength is the efficiently scanning of large amounts of data across and large number of investment managers. Whereas humans can provide the nuanced analysis required to make the final investment decision.
Using a human / machine hybrid approach a machine learning algorithm could be trained to run an initial screen of investment manager universe. Essentially acting as an idea generator.
The algorithms would be trained with data sets and use algorithms to generate a subset of investment managers most likely to generate alpha.
This subset would then be passed to the portfolio managers and analysts for further review, manager meetings etc. and a final decision made.
Example: Potential use of ML in investment manager selection
Like anything new and unfamiliar it takes time for the investment community to adapt and become more comfortable with the new technologies. Implementing this approach would see machine learning as an addition to the manager selection process rather than replacing the process.
It is Shorelines belief that as Asset Owners see the benefits that predictive analytics can bring to their investment approaches the technology will become more common place.
Shoreline has a cross section of knowledge in AI, Investment Management, Operations, and Data and Analytics provides the necessary experience required to work with asset managers and owners to identify suitable problems, develop, and operationalise AI tools. We are seeking a partner to work with us in a pilot proof of concept. Please contact us if you have an interest in participating.
Chris Robertson
Associate Director