CEO explains what a generative mathematical model can do to help clients outperform the market
Each month at WP we offer a slate of articles and content pieces that go deep on a particular topic. This November, we’re exploring the uses of artificial intelligence in wealth management.
Advisors across North America are stuck between a rock and a hard place. Fee pressure and low-cost brokerages force advisors to justify their value. The shift from a distribution to a service business has imposed new expectations on advisors who now have to weigh in on questions far removed from investment management. The investment management side has grown more sophisticated, too, and clients will gauge their advisors against index performance they could get for less money. The answer from firms has been to better personalize service delivery, attracting more complex and higher net worth clients who need personalized service. But scaling a practice and going deeper with clients pull from the same finite pools of advisors’ time and energy.
Technology is often touted as the answer to this apparent dichotomy between scale and personalization. But Christopher Ainsworth argues that tech adoption can actually add complexity to advisors’ work, at least in the short-term. The CEO of Pave Finance, a US-based AI wealth management software company, explained that as the range of services advisors are being asked to deliver widens out, adding new technologies creates a situation where they end up with “junk on top of junk.” He argues that where tech can add value is in the personalization and scaling of investment management, something that he believes his firm can help advisors achieve.
“The advisor has to constantly be migrating their business upscale. So they've got to take a certain group of clients when they start their business and then migrate upscale to make themselves more profitable. Because the one thing we can't create more of is time,” Ainsworth says. “From there, they have to figure out how to manage the small asset base of their book. 30 per cent of an advisor's time is spent to generate less than 10 per cent of their revenue. And so the biggest single challenge is, how do they make that 30 per cent of their time more profitable? And so that is something that we at Pave have sought to solve.”
The solution Ainsworth offers is a combination of tech automation and generative artificial intelligence. Pave’s software is cloud-based, and connected to advisors’ brokerage firms. The software, he says, never touches client assets but it does have access to client information. It allows advisors to input clients’ objectives, risk tolerances, and other investment preferences. From there an advisor can select a benchmark with their clients, be that an equity index or a multi-asset strategy. The portfolio will then be monitored by a convex multi-scale optimization engine that will seek to maintain the portfolio’s adherence to that benchmark from a volatility standpoint. It immediately creates a touchstone by which advisors and clients can assess portfolio performance against volatility and risk objectives.
The most important choice that the software gives clients, Ainsworth explains, is whether they want a passively or actively managed strategy. A passive strategy will be maintained through automation, with direct indexing, tax-loss harvesting, and appropriate rebalancing, keeping as close to the chosen benchmark as possible while allowing for those personal investment preferences. When a client elects for an actively managed portfolio, Pave’s AI steps in.
The AI model used at Pave, Ainsworth explains, uses predictive analytics and machine learning like a large language model (LLM) but unlike an LLM, the model is trained on numbers rather than language. The model has been trained on hedge fund quantitative models, allowing it to predict what could happen with any security on any exchange in the world. The AI model will then pick stocks to replicate the chosen benchmark’s level of volatility while outperforming that benchmark on a returns basis. Ainsworth says that the models this AI tool is trained on have a track record going back to 2010 and have delivered roughly three per cent outperformance against benchmarks per year. Ainsworth argues that this approach allows for a client’s personal needs and desires to be integrated while driving returns with minimal effort on the advisor’s part.
That is not to say Ainsworth believes this model can manage client money alone. He notes that every quantitative model ever used without appropriate human oversight has eventually failed. The assumptions written into quant models can be upended, as with the 2008 financial crisis. Issues like liquidity are often unaccounted for, as well as the sheer volume of trades proposed. To prevent these issues, Ainsworth explains, the Pave software goes through a series of checks on a weekly basis and each stage of the process requires human review and approval, retaining advisor control.
Data security, too, is a paramount concern. Ainsworth emphasizes the proprietary and segmented nature of his firm’s operations, as well as the fact that the AI never touches client assets, it merely observes and makes suggestions.
While the Pave software system is not yet available in Canada, Ainsworth notes that his firm is exploring the possibility of expanding north of the border. Whether its his model or another model that ends up in Canadian advisors’ toolkits, however, he stresses the argument that models like this will change some of the dynamics currently pressuring advisors.
“It's a natural fit for us to go north of the border. It's a natural fit for us to cross the Atlantic to Western Europe,” Ainsworth says. “There's common, common language, there's common legal structure, there's common infrastructure for all of us. So those are places that we're already looking at and looking to provide the software out into those markets to those advisors through larger organizations.”
