It is not simple being a wealth manager in 2018. Regulatory overheads, pressure on margins, the implicit complexity of keeping up to date with the latest technology while still needing to run complex change programs, all add to the headaches of the modern bank executive.

Here, I present some techniques that we have used to help banks to diagnose and treat the migraine. A modern wealth advisory process is commonly broken down into logical stages:

• Understanding the client’s situation. Here the advisor performs a process to measure the risk tolerance of the client (how much risk the client prefers to take) and the risk capacity of the client (how much risk the client’s circumstances allow them to take).
• Defining a suitable investment strategy. The advisor may recommend that the client’s wealth tracks a portfolio with certain risk or reward characteristics, or follows a certain structured product strategy.
• Recommending suitable investments. The client is presented with a portfolio that meets the investment strategy.
• Controlling and monitoring the client’s wealth portfolio. The advisor shall periodically monitor the wealth portfolio to ensure that it continues to meet the investment strategy and the client’s situation.

Regulations such as MiFID 2 (the Markets in Financial Instruments Directive) are designed to ensure that the wealth manager proves that investments recommended are suitable for the client; therefore, firms have no choice but to implement repeatable processes and record evidence.

Digital internet-based technologies allow these processes to be implemented reliably and flexibly, which thus enables other value-added services such as online capability, mobile usage, increased customization or alternative interfaces such as video or chat.

Let’s go through the advisory process stages in more detail, and see how a systematic quantitative approach can help to bring coherence as clients move through the process.

First the client profiling: the problem is to describe the client’s risk preference and client circumstances as one or more numbers. Two approaches are commonly used – ask the client “Where are you on this risk scale?” or give the client a questionnaire and infer the risk preference from the answers using a model. The problem with the former is that often the client cannot easily answer the question, and the problem with the latter is that the data required to validate the model is not available, and therefore it is based on heuristics. Estimating risk capacity is also problematic, as the advisor may not have all the data in the case of held-away investments. Could secure application programming interface (API) integration with custody institutions hold the solution?

Whatever the approach taken, digital technologies help to differentiate this stage – immersive design, playing an economic game to measure risk appetite, or mining client profiles or transaction data can all be considered.

Now we have the client risk preference, is it possible to automate the selection of an investment strategy? In the past, this would have involved a discussion with the advisor: what is the right account type – advisory account, execution only or investment mandate, or a combination of all three? This decision can easily be captured over a digital channel. For execution-only, the client demands information, research and crisp execution, which has been available in digital form for decades. With investment mandates, the client is looking for a hands-off solution but may still specify preferences and restrictions, for example ethical investments only. The business model is about efficiency, so only clients with high assets under management were offered restrictions, as this required specialist investment management. Now quantitative software solutions are available that allow the efficient implementation of restrictions for the mass market, thus there is a convergence of advisory and investment mandate management.

For standard advisory, the client risk preference would be described on a discrete scale, for example 1 (low risk) to 7 (high risk); it was trivial, therefore, to map clients to a set of seven matching portfolios selected by the CIO of the bank. The issue with this implementation is that neither is client customization taken into account, nor is it easy to make an argument that the entire variety of client circumstances, tolerance and aspirations can easily be mapped to the fixed set of portfolios. For such implementations, it is harder and harder for institutions to justify the fee premium over competing solutions such as exchange-traded funds. Is it possible to create a model that is flexible enough to describe client customizations and variety, but also systematic enough to be implemented cost effectively? We believe that it is, by using techniques such as portfolio optimization and goal-based investing. The (lack of) performance of some CIO strategies can also be mitigated by allowing the client to specify which benchmark portfolio they would like to track (perhaps from other investment managers).

For other advisory clients that are not interested in a portfolio-based approach, the bank needs to provide a way to generate, propose and implement ideas. Traditional approaches to idea generation (such as macro or thematic investment) can be combined with more fashionable approaches such as using big data management techniques in combination with machine learning models. Time will tell on the feasibility and future returns. Any ideas proposed need to be suitable with respect to the client’s profile; therefore, the advisor needs to also measure the risk of the strategy/idea. This requires a sophisticated risk management approach – see Figure 1.

Once a strategy is chosen, is it possible to automate the process of selecting investments? This depends on the exact strategy, but in general, yes.

For a benchmark tracking strategy with customizations or restrictions, portfolio optimizers can be used to generate a new portfolio. The optimizer can take the benchmark portfolio as an input and minimize the tracking error to it, yet still be faithful to the restrictions. For idea generation strategies, it is possible to create solutions that search the available universe of products for the best match to the strategy. Additionally, it is possible to search the universe of products that can best be added to a source portfolio for hedging or other reasons.

Digital technologies allow easy large-scale monitoring of client portfolios. Some monitoring checks are simple and data-driven, for example portfolio concentration risk, but some checks require sophisticated risk management techniques, for example calculating a risk measure of the portfolio is within defined bounds.

The cloud enables the scaling up of monitoring to hundreds of thousands of portfolios, by providing almost limitless and elastic compute, yet the choice of algorithm is also hugely important for an efficient risk management solution. Consider a typical wealth manager’s investment offering of 100,000 products. In order to calculate the risk of the portfolio in aggregate, the bank must analyse which products have diversification effects with others. This presents a problem with high dimensionality – 10 billion combinations need to be analysed! Our recommendation is to instead model the investment world using risk factors – a set of asset classes/markets/regions/industries that describe the majority of the investment risk. The 100,000 products are then mapped to the risk factors and only the diversification effects of the risk factors need to be modelled. This enables the portfolio risk to be calculated quickly, which is especially important if the risk management needs to be exposed over a digital channel or if what-if analyses are taking place during an advisory meeting.

So, it is possible to automate the process from the beginning to the end and provide robust and reproducible advisory processes! This capacity for automation then enables either fully automated or hybrid advisor business models.

Even though the business model can be automated, digital solutions present wealth managers with many management challenges.

A common approach is for the four functions to be abstracted behind robust and secure APIs, which enable the bank delivery channels to innovate without changing the business engineering and processes behind them. These APIs have essential complexity in terms of data security, access control, API management and versioning, self-hosting vs cloud. The mastering of this complexity is often jurisdiction and institution specific, so general guidelines cannot be given other than to say that solving these details requires expert advice and should not be underestimated.

In summary, the intersection of increased automation and adoption of digital technologies leads to many challenges that can be successfully (but not easily) solved using quantitative techniques, innovation and investment in technology, and capable change management.