This is a Guest Blog by Anu Khanchandani Co-Founder & Principal Mentor (IT), Fintoo
Being a wealth manager is a tough job.
More so, now that there is a plethora of data that is to be mined before one can come up with an ideal asset allocation for their clients. Not only that, real time wealth advice which is in tune with the dynamically changing markets can add value which can keep your client with you for the long term.
Also, when that real time advice is sought by the client in an inbound mode because you were intelligent enough to raise curiosity by sending them triggers to take action on their portfolio, it adds the much desired icing to the cake.
This is the kind of ideal situation that every asset manager vies for, however, knows that is hard to attain.
The research that is required to bring a client to that point where you propose an ideal asset allocation which snugly maps to their goals is the first and most difficult step. The second is of course executing the proposed asset allocation deftly. Last but not least, the activity of continuously reviewing their portfolio to send appropriate triggers to keep the stickiness on is daunting.
It is clear that technology has to come to the rescue to make all this seamless and comfortable for the asset manager, who can in turn make their clients’ lives a lot easier by ensuring that they are given real time wealth advice at all times.
Enter Natural Language Processing (NLP).
What is Natural Language Processing?
Natural Language Processing (NLP) is a subset of AI (Artificial Intelligence) which deals with nothing but text crunching. Why is the financial services industry suddenly raving about text crunching? Didn’t we always think that this is the industry which played with numbers more than with text? Here’s the thing. Till some time back, clients were satisfied with portfolios with proposed asset allocation of companies based on what the numbers in the companies’ annual report said. However, now there is the possibility of getting better insights using news and other text material like regulatory news bits, analysis reports, e-commerce activity, talks, videos, geolocation data, and satellite captures. Basically, any information that can be linked to an industry, company, or general economic movement has potential to be source data for the wealth manager to allocate the right mix in their client’s portfolio.
This kind of data is what we call unstructured data. Numbers are pretty structured and can be processed and understood in a simple tool like an Excel spreadsheet. However, processing unstructured data, whose format is not defined, is a challenge which Natural Language Processing can seamlessly resolve.
Natural Language Processing, can crunch this humongous amount of unstructured data at speeds that no human can. If used effectively by wealth managers, this could be a game changer in the way that they give advice.
3 stages in which NLP can help Wealth Managers
Pre Investment Stage
The pre investment stage for a wealth manager is all about analysis of the market and the client’s goals. The deliverable is an optimal asset allocation which maps one-on-one with the client’s defined goals.
Without NLP, the scenario looks like this – Multiple research analysts spend days and night analyzing company reports and other statistics and raw data. All this is done manually, in the sense, there are actually people reading this data and drawing inferences and noting them down.
Here’s what hurts – Out of all the analysis done, hardly 20-30% of all the reports read contribute to the final recommendations of asset allocation mapping against goals.
NLP can take up most of the manual work for a wealth manager by dissecting and merging structured and unstructured data, forming patterns, and assigning weightages to the discovered research points. Those research elements on highest priority can be used by a human advisor to make better and faster decisions than going through all the data manually, discarding what is irrelevant and then focusing on just the relevant part of it.
Investment Stage
Once the pre investment research phase is done, the client has obtained their deliverable of a financial plan which has the optimum asset allocation for their goals. Now comes the stage where it is time to bring the plan into action.
An asset allocation needs to be made actionable by actually advising the client on the actual instruments that need to be invested in.
In this stage NLP plays an important role by processing the combined input and decision data to produce a standard unbiased report. This report, in addition to providing recommendations, also explains the rationale behind the same. The explanation justifies the recommendations and the potential contrary factors.
Portfolio managers can then use this intelligent report to review and approve or reject the investment recommendations.
Post Investment Stage
The post investment stage is the critical point where stickiness inducing activities like portfolio review and rebalancing can keep the client coming back for more. This kind of inbound reach from the client can create healthy client-advisor relationships for the long term.
In this case, NLP can be used in 2 phases.
The first is the review phase where instead of a human advisor going through the portfolio in detail, NLP can be used to analyze the current portfolio programmatically to give insights into deviation of recommendation WRT the client’s goals and raise alerts where the deviation goes beyond a particular limit.
The second is the rebalancing phase, where NLP performs the same activity that it did in the investment phase to find suitable investments that will bring back the goal achievement plans on track.
Use-Cases of NLP in Wealth Advice
There has been an increasing demand from institutional and individual investors alike to factor in ESG (Environmental, Social and Governance) data while researching an optimal asset allocation mix for a portfolio. Deutsche Bank had undertaken a project to filter out companies who reported ESG activities in their reports just for the heck of it. Using NLP, the team at Deutsche Bank shortlisted only those companies which actually used words related to mitigation and adaptation rather than just reporting risks, footprint and monitoring of sustainability measures.
JPMorgan Chase used the Feature Selection aspect of NLP to review commercial loan agreements. Basically, it involves using groups of words belonging to a certain category and assessing if any of these sounds risky and deserves more attention from a human expert. The bank reported that the solution saves lawyers and loan officers work 360,000 hours annually.
N26, a leading bank in Europe, has employed a NLP based chatbot to service its huge customer base of 2 million. Not only did it help provide effective customer service in spite of rapid scaling, but also improved customer experience and operational efficiency, through faster responses to customer service chats. Soon after going live in the mobile app, N26 quickly saw 20% of customer service requests handled by the AI assistant. N26 is working on bringing this to 30% and beyond.
Closing Thoughts
Wealth management has been and always will be a human driven domain, primarily because it deals with people’s money. Humans will definitely do a better job at comprehending high-level written text which adheres to a particular context. However, with the rise of a plethora of sources of unstructured textual data, it makes more sense to leverage the advantages of technology to process such data.
Advances in AI and its subset NLP, which are blessed with humongous computational power, are here to change the way that wealth managers can process data to give better and better recommendations. There is a huge divide between the firms that leverage technology and data for a competitive edge, and those who are falling behind.
At the end of the day, it is a simple choice between hard work and smart work.
