Wealth through Investing

“Augmented Intelligence”: Combining Human Intelligence and Technology

[ad_1]

It seems like artificial intelligence (AI) applications are showing up everywhere. The development of self-driving cars is advancing rapidly. Netflix analyzes your viewing habits to suggest shows you’ll like, and Spotify does the same for your audio playlists. Google has improved its spreadsheets with AI features that can respond to phrases as inputs versus formulas. AI is also making inroads into private wealth management in both investment management and practice management. It’s an exciting time for AI, but will it cause more disruption or innovation for wealth managers?

New Ways of Thinking

AI has been around for years, but the pace of innovation and adoption has increased recently. Anand Rao, partner and innovation lead with PwC in Boston, cites three primary reasons: (1) increased computer power to run AI algorithms; (2) increased availability of data in sufficient quantities to train AI systems; and (3) increased use of cloud-based, centralized storage for large datasets, which also facilitates training.

The term AI includes several different approaches, including machine learning, natural language processing, and neural networks; the discrepancies between these have led to some confusion about how the technology is really defined. Edgar van Tuyll, chief quantitative strategist at Banque Pictet & Cie in Geneva, Switzerland, used the following definition of machine learning to explain it to me in an email:

“A precise but technical definition of machine learning could be: statistical (based on probabilities, not certainties, so users have to accept a degree of error, which is something no one talks about) prediction algorithms (completely automated) that change with iterations on new data (the parameters or the algorithm itself change based on new data) without human intervention and minimize a cost function (the cost function is some function of the difference between predicted and actual values).”

This approach differs from automation, he adds, which is a product of exponentially increasing processing power and data but does not use algorithms that evolve.

Rob Stanich, global wealth management offering manager with IBM Watson Financial Services in New York City, prefers to define AI as “augmented intelligence” to highlight the human interaction at the process’s core. He cites machine learning and natural language processing as two AI technologies upon which IBM is focusing. Natural language processing is “the ability to understand human language, to draw conclusions around meanings, semantics, sentiments,” Stanich explains. “We have technologies within that space like personality profiling based on writing sample, tone analysis, whether somebody is angry or scared when they’re speaking or writing.”

Quant-ification

Not surprisingly, some quantitative investment managers have been early AI adopters. Steve Wilcockson, industry manager for financial services with software developer MathWorks in Cambridge, England, cites the example of an asset manager who is using “machine learning to determine correlation and predictive trends across macroeconomic, credit, liquidity, risk, and money-flow factors. This allows them to better understand asset-class performance trends against risk, with some of their portfolios outperforming benchmarks by 100 basis points,” he told me via email. Recurrent neural networks, which loop previously gathered information to better provide context for handling new data, can be useful for volatility estimation, while classification and regression trees have proven useful for back-testing (with foreign-exchange strategies, for example).

Stan Sakar, founder and president of Abaris Investment Management in Detroit, Michigan, uses a neural network that took more than 10 years to develop. Neural networks mimic the human brain, he explains, in that they receive inputs from multiple sources, analyze those inputs, and compare them with previous observations and outcomes. That process allows the network to learn and adapt its forecasts in an iterative fashion.

The network’s output is a buy, sell, or hold indicator for the asset, and about 90%–95% of recommended trades are automated, with the balance subject to manager discretion. It’s a highly complex system. Although he doesn’t have a precise development cost, he estimates it’s in the $10 million to $15 million range.

As Sakar’s experience indicates, successfully incorporating AI into investment management is a significant challenge — this is not plug-and-play technology. Van Tuyll shares that perspective. “What works is combining human experts and machines. Using scikit (popular Python code libraries for machine learning) blindly on data you do not understand is a path to failure,” he says. “And continuing to run investment like before machine learning is like sticking with horse-drawn carriages in the age of the first automobiles: Your driving skills will at some point not compensate for the car’s technology, but you, driving a car, will be unbeatable.”

AI in Practice Management

The use of AI for practice management could speed its adoption among wealth managers. Salesforce recently announced its Einstein platform, which is intended to enhance three key activities, according to Rohit Mahna, senior vice president of financial services with the firm. The first is increasing productivity. Einstein will take over automated tasks and help advisers become more proactive with their data usage. Mahna gives the example of a computer dashboard that makes suggestions on who the adviser should call that day.

Uncovering business opportunities by mining the adviser’s existing book of business is the second activity. “How can you intelligently uncover new opportunities based on prospects or existing client sentiments, maybe potential competitor mentions or overall prospect engagement that you’ve had with clients over time?” asks Mahna. Einstein works in the background to look for patterns that indicate opportunities with existing clients or prospects

Retaining clients is the third activity. Einstein monitors client-related activities to spot patterns that indicate which clients are at risk for leaving the adviser. IBM’s Watson Client Insight for Wealth Management solution provides a similar forecast. “We’re able to predict with a high degree of accuracy when clients are going to leave the firm — 30, 60, 90 days ahead of time,” says Stanich.

Smaller firms are also integrating AI into private wealth management. ForwardLane in London and New York City is combining quantitative investment models and financial planning. Responsive Capital Management in Vancouver offers its Alpha Digital Advisor Platform and AI Research Platform. And New York City-based Synechron has developed its Neo suite of AI services for financial advisers, which includes natural language processing, chatbots, machine learning, robo-advisers, and other services.

All About the Benjamin

Wealth managers frequently cite the difficulty with taking on clients who don’t meet their minimum asset requirements, such as the adult children of existing clients. Matt Reiner, CFA, CEO and co-founder of Wela Strategies in Atlanta, Georgia, uses AI to attract and service that market profitably. Reiner started his career with Capital Investment Advisors, a traditional wealth management firm his father founded. The desire to serve their less affluent clients more efficiently led Reiner to start Wela, an online, AI-based financial coaching service aimed at young families.

The goal with Wela was to digitally recreate a financial adviser’s intuition by engaging users with a digital personality named Benjamin. “Benjamin is an AI thought-like technology that analyzes situations,” Reiner explains. “It’s really the middle ground between the user and our financial advisers here.”

The system, which the firm developed internally, aggregates users’ financial information and generates a daily spending limit based on their free cash flow and savings goals. Reiner describes it as being “like Weight Watchers for budgeting.” Users can contact human advisers as needed and invest in Wela’s model exchange-traded fund portfolios, which serve as the income source for the service.

The Back in the Future

According to a recent white paper from Capco titled “Transformative Nature of Artificial Intelligence (AI) in Wealth Management,” wealth managers have been slower than other financial services segments to adopt AI. Nonetheless, the paper’s authors believe AI will have a broad impact on wealth managers’ information flows and organizational structures.

Gary Teelucksingh, a Toronto–based Capco partner and co-author of the report, says that clients, advisers, and investment managers currently view specific types of information at specific access points. Focused, well-entrenched technology providers service each segment and its data flow.

That segmentation is breaking down, however. Clients are increasingly requesting and gaining access to more-sophisticated information about their portfolios. Advisers and investment managers are using AI and other big data capabilities in more sophisticated ways to develop their advice. Essentially, the three streams are now coming together as one, according to Teelucksingh.

This change can’t be attributed solely to AI. Instead, the AI component is joining with the ongoing trends of digitization and automated workflow. Data that were traditionally stored on paper have been digitized as the financial services industry moves to digital delivery. Those digitized sources can then be interrogated and their data pulled from them so other systems can work with the data. Consequently, workflow processes exist today that allow firms’ back offices to function more efficiently.

Those changes will directly affect operations because once the data are available employees aren’t necessarily required to interrogate the data. The result of this transition will be increased efficiency and lower costs. “The way in which it’s going to happen is the roles that are not client-facing today are the ones that are going to be most highly impacted, because they can be driven by confidence-based decisions,” he says. “Look at this, look at this, draw a conclusion. If your confidence level is 99.6%, take action. These things are no different from how the back office works today — those functions [as performed by humans] are going to go away.”

Will the trend to greater back-office automation mean the demise of humans in front-office roles? Teelucksingh doesn’t see that happening, because the advisory and trust-based elements of the firm’s services will still be primary, and those elements are best delivered by other people. The technology will cause employees’ roles to evolve, however.

“The investment manager will actually become highly specialized at model tuning, model origination, model management,” he says. “The adviser will become more attuned to using the outcomes of the data to help proactively service the client. So, the end result is probably 50% or more reduction in cost by way of staff, probably somewhere similar in nature in percentage reduction in cost to service the customer, which will include technology, and that should in fact drive up margins.”

This article originally ran in the September 2017 issue of CFA Institute Magazine.

If you liked this post, don’t forget to subscribe to the Enterprising Investor.


All posts are the opinion of the author. As such, they should not be construed as investment advice, nor do the opinions expressed necessarily reflect the views of CFA Institute or the author’s employer.

Image credit: ©Getty Images/ R_Type

Ed McCarthy

Ed McCarthy is an author and freelance finance writer who covered private wealth topics for CFA Institute Magazine. He has written for many of the financial service industry’s leading publications, including Bloomberg Wealth Manager, Institutional Investor online, Financial Planning, Journal of Accountancy, and the Journal of Financial Planning.

[ad_2]

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *