Menu Close

Wealth Management Is a Tech Late Bloomer, But AI Will Change That

view original post

By Joseph Wang, Junior Data Scientist at VRGL 

The wealth management business has traditionally been late to adopt new technology, primarily because of caution about security. It took nearly a decade for firms to trust putting data in the cloud. Today, many other industries are leveraging artificial intelligence (AI) to improve processes. The wealth management industry is still using processes that are tedious and manual. But that is about to change.

AI will not only accelerate but automate both client acquisition and service in the wealth management space. Acquiring clients is very relationship-driven, often consisting of numerous dinners and rounds of golf.  Today, when a relationship between the prospective client and wealth manager is established, the wealth manager will begin the manual process of entering the clients statement data into analytics software to become familiar with the portfolio.  After that, a proposal is generated to bring forth ways to improve the current portfolio.

Unlock hidden potential 

Once a decision is made to hire the wealth manager, the process begins of maintaining the ongoing relationship, managing the portfolio, trading, compliance, among other responsibilities.  Described above is an example of the traditional prospecting journey for wealth managers. It can take several months to even years as it is challenging to build trust and time intensive to extract, process, and analyze statement data manually. How can AI accelerate this process and simplify decision-making for wealth managers?

In addition to all the process improvements AI can bring to wealth managers, it can also simplify decisions in both client acquisition and investment management. As mentioned above, the client acquisition process for a wealth manager can be long and arduous, but AI can provide solutions as to whether a prospect should be pursued at all. Using AI, a recommender system can be built to suggest wealth managers to clients based on the client’s risk tolerance, ESG beliefs, and other characteristics.

Recommender systems have been in use for decades and are how Google suggests ads to you based on your search history or how Netflix highlights movies you may want to watch based on your history. Based on a client’s past investments and current characteristics, a recommender system could provide a list of recommended wealth managers who are suitable. Not only does this save the prospect time in finding an appropriate wealth manager, but also allows the wealth manager to save time not pursuing incompatible prospects.

Once hired by the client, the wealth manager often selects an investment strategy for the client in the form of a model portfolio based on the client’s characteristics. Using AI for this selection allows the wealth manager to be more confident that the decision is backed by quantitative reasoning. Many unsupervised learning methods, such as clustering or expectation-maximizing (EM) algorithms, can find patterns in data and group similar data points. By applying such methods to a group of clients, wealth managers would not only be able to better understand similar clients, but also quickly and confidently categorize a new client. By selecting model portfolios using unsupervised learning, wealth managers would reduce the time analyzing client portfolio, as well as be confident in the quantitative validity of the result.

Combatting challenges

AI implementation in the wealth management business is not without challenges. The requirement for data accuracy is a continued pain point in statement parsing. Tuning the confidence level required for OOD data (Out-of-Distribution; that is, data or formats never seen before) would be challenging and may pose the risk of leaking confidential client information.

Privacy and security become a top priority when utilizing AI, and wealth management firms should develop their internal framework and preemptively address client concerns about the use of sensitive information. Additionally, even with significant inroad, decisions made by AI are still relatively poorly explainable, sometimes even unreliable when facing new data, and the regulatory landscape is constantly evolving. Practitioners are best served using a mosaic approach, retain a healthy suspicion for AI-provided recommendations as to any other source of information, and be ready to intervene.

Heightened efficiency with AI

AI has already positively affected many industries from automating tedious and manual tasks to bringing to light new information in formerly collected data. As AI solutions continue to be adopted by wealth managers, it will only become clearer just how much more efficient the process can be.