There are several recurring topics on the agenda of the modern wealth management industry. Notably, most of them refer to digitalization. The keynote speakers at wealth conferences deliver presentations on whether full-digital or hybrid models would be preferable. Panel experts discuss the pros and cons of shifting the operations to the cloud, and the industry contemplates the usefulness of chatbots to increase customer engagement and brand awareness in their digital channels.  

Amidst the strategic decisions and the fears of a mysterious AI stealing the jobs of financial advisors, we believe one important detail remains overlooked. Do we properly understand the machines that are to automate an essential part of our value chain or that may become an alternative to our human operators?

Do we assess them with the same scrutiny as for a new employee we would hire for a job?

Although we tend to classify people for various purposes, there are hardly two mortal financial analysts thinking the same way. Human judgement is composed of multiple convictions; it is affected by individual biases, personal quirks and mood changes. In theory, there is an infinite number of mechanisms suited for automated investment decision-making, but there are two distinct types that we repeatedly come across in this space. Let us review each one of them in greater detail and discuss their importance for the long-term outcomes of digitalization projects.

The first and the most common principle of automated investment decision-making is Modern Portfolio Theory, which earned a Nobel Prize in Economics for Harry Markowitz. Since then, it has become the first topic in any introductory finance curriculum around the world. The machines based on this principle usually estimate the mean and variance of the assets’ returns based on their past performance. Then, they estimate correlations between these returns and calculate portfolio weights for all possible combination of assets. After that, the machines maximize the expected return for a given standard deviation over the set of weights. Finally, they determine a suitable portfolio, given the investor’s level of risk aversion.

Modern Portfolio Theory, in its most popular form, suffers from several drawbacks. First, these machines are single period by design. This means that they estimate their parameters from historical data for a given time horizon. This feature makes it impossible to include features such as rebalancing or realistic tax deduction, where monthly account balances can differ for each year. Second, the assumption that returns follow a normal distribution is unrealistic. It tends to underestimate large losses and thus give too optimistic predictions. Third, this approach is siloed in its nature, because it is difficult to use it for models assets and debts together or include more complex instruments such as mortgages in the analysis.

What is the alternative?

Another family of models able to automate financial decision-making is far more sophisticated and elegant. The scenario-based models adopt an approach that allows carrying the entire balance sheet of the end-customer into the future, cashflows included. These machines allow for more advanced probabilistic modelling of the underlying assets and debts on the end-customer balance sheets. The scenarios fueling these machines can easily incorporate wealth managers’ outlooks on how economies will develop into the future. Like in the case of the first family of machines, the optimal portfolio for a given customer is determined by maximizing a utility function, defined by their unique financial situation and risk tolerance.

There are numerous distinctive advantages of the scenario-based approach of financial machines over the standard, simplified use cases of the modern portfolio theory.

First, scenario-based models are multiperiod, which allows financial brands to differentiate themselves by providing additional functionalities such as automatic rebalancing and realistic tax deduction, which would not be possible under the first approach.

Second, unconstrained by the assumptions of normally distributed returns and the quadratic nature of the utility function, the scenario-based machines allow for more accurate stochastic modelling of institutions’ product universe. This would have a similar effect on the resulting calculations of consumers’ finances as would an increase in your phone’s camera resolution on the pictures you’re taking.

Third, amplified by the cloud, these machines are no longer constrained by poor performance but can evaluate thousands of realistic scenarios per second, contributing to a fast and seamless customer experience, which is critical in a digital context.

And fourth, the full balance sheet approach allows modelling of virtually any financial situation that the end-customer could face throughout their lives. This design supports a number of customer journeys only constrained by the fantasy of the financial institution. Essentially, this technology allows for building out a genuine family office offering for the masses, destroying siloes in digital wealth once and for all.

So, ask yourself again, which new employee would you hire for your department? An undergraduate that has an ambition only ever to be familiar with modern portfolio theory; or a flexible grad-school level professional skilled in stochastic modelling? The professional that with your guidance and assistance, can model any complicated situation of your customer?

You might wonder, given such a dramatic difference at the very heart of these machines, why many reputable institutions carry on cutting corners and spending millions on developing inferior experiences for their customers. We believe the answer is in the short-term strategic outlooks and a reactive digitalization strategy pursued by many such institutions. The user experiences might look similar, but the first type of machines heavily limits an opportunity to develop a holistic approach to consumer’s finances. Therefore, these companies risk having to replace a newly built, expensive customer journey with another one in the future.

Although the FinTech space tends to criticize the incumbents for a myopic attitude to digitalization, we have an example of the opposite. Some established financial institutions adopt a surprisingly hands-on approach to the digitalization of their businesses. Take Skandia Life, one of the largest life insurers in Sweden, who have recently released a hybrid pension planning solution. Since July, Skandia’s customers can start saving for their pensions in a channel of their choice – be it a physical meeting with a financial advisor or a digital journey – all driven by a consistent, scenario-based machine at the heart of the solution. Currently, this approach helps Skandia own the seamless and fast experience of the solution. In the future, it would enable the insurer to easily extend their offering to support more wealth journeys, laying a path for pioneering open finance with a holistic view of their customers’ finances. 

The quality and attention to detail are not always visible or understood by the customers of the financial institutions. Although many banks and insurers made commitments to digital strategies, there is still a lack of discussion about which machines would replace which human tasks and how good they are. We believe that choosing a superior technology to drive your digital solutions could be not only a reliable differentiating factor to your brand but also a great strategic choice for enabling further opportunities to innovate. Given the right technology, it is possible to give the prediction machine the soul, the spirit and the story of your financial institution. And most importantly, the right machines are better suited to learn, develop and adapt to your vision as times change.