When one thinks of Artificial Intelligence or complex computer models, typically the mind leaps to talking robots or driverless cars. But the reality is much closer to home and the applications in the world of finance are game changers. Tasks that would have required millions of man hours prior to the evolution of computing power, are now possible in hours and one of the most powerful areas where technology can help is in the domain of fund selection and portfolio management.
Back in the dark days of financial advice, fund selection was often driven by hearing a manager speak that you like or picking a fund management group that had a good name for an asset class; M&G for bonds, Baillie Gifford for Japan and so on. Or it might be from a cursory glance at Money Marketing in the sector tables to find the top quartile funds in bold. More recently, and prior to recent developments in fund screening, an experienced research team would decide that the best way to create a buy list was to look at discrete returns over the last few years and come up with a ranking for each sector. They might even dig a little deeper and see how the returns had been achieved by looking at returns adjusted for volatility, or even establish if the manager was achieving returns through beta or alpha. But that would likely be the end of the analytical journey, concluding that their methodology was applicable to each sector in the same fashion.
However, if you wanted them to look at all the k ey performance characteristics of all the 4000 FCA regulated onshore funds, including alpha, beta, sharpe, short term performance, volatility, drawdown etc… over the last few years it would require enormous resource. Particularly if you wanted them to establish which characteristic was most crucial for relative outperformance in each sector and then back-test to see what combinations identified the funds that delivered the result you were after. However, thanks to the advancements in data availability, some Clever maths and the grunt of cloud computing it is now possible to process this information in a matter of hours.
With thanks to Steve Nelson at the Lang Cat who wrote in the Q1 issue of this very publication, predicting; "Clever Algorithms from Clever People" delivering enhancements to current investments practices. Apart from being spookily "on-brand" (Steve and the lang cat are not affiliated with Clever) he acknowledges that one of the few risk-free predictions in our industry is that the speed and complexity of technology will increase and (in theory) make all our lives easier.
By building complex 'what if' algorithms that test and re-test millions of data points, they are able to create a bespoke fund selection for each sector that not only provides a ranking system for buying funds, it crucially also tells you when to sell them.
Clever Technology Ltd. have built the necessary platform to complete such a task. By building complex 'what if' algorithms that test and re-test millions of data points, they are able to create a bespoke fund selection for each sector that not only provides a ranking system for buying funds, it crucially also tells you when to sell them. Critically, the system keeps learning. Every month the Clever algorithms are refined as new data becomes available, so the criteria for fund selection is constantly evolving – try asking a research team to rework their process every month! This approach not only works in theory but has delivered quantifiably superior fund selection and monitoring that has powered adviser portfolios in the real world for nearly 10 years.
The question is then how to translate this superior fund selection in to robust, risk graded models, and this is where Clever lean on the experience of the 8AM Global Investment Management Team and the risk profiling expertise of Dynamic Planner. Essentially, the Clever-driven fund selection is mapped by 8AM Global onto Distribution Technology's Asset Allocation and adjusted reactively in order to comply with their Risk Target Managed profile. This Dynamic Planner product guarantees each models continuous adherence to its risk profile rather than just at quarterly observation points.
Many model portfolios and multi-asset funds are run with specific exposure 'budgets' for each sector exposure, which are then populated with funds that fit the risk and return characteristics required to fit within those given allocations. The Clever investment process turns this on its head, as fund selection becomes the fixed point around which the asset allocation must flex to achieve consistent risk exposure.
This choice is borne of the desire of the whole team to avoid human bias wherever possible, and that if we believe that the funds picked by the system are quantifiably going to yield the best result – why dilute that by picking a lesser fund that fits the current allocation without weighting adjustment? Or colour the asset allocation with our own views, which irrespective of experience or expertise adds often needless complexity to long-term strategies.
This approach to fund selection and the building of model portfolios, is simply not possible without human input at key stages, but it does take away the 'heavy lifting' of data crunching and the unconscious (or conscious) biases that comes with humans performing this part of the fund research process. It also provides a compliant and repeatable framework for fund selection and quantifiable, auditable justification of any portfolio changes made.
If one was ever in doubt of the benefits that the application of technology can bring to this evolving world, IBMs Watson has already demonstrated a far more accurate diagnosis rate for lung cancers than Dr's can achieve, and my car is far more competent at parallel parking than I could ever hope to be.
In the ever-crowded world of model portfolio offerings, the 8AM Clever Models stand apart as evolutionary; using the technologies of tomorrow to provide a fair and transparent methodology by which to manage client assets.