Traditional asset managers are increasingly applying data science and advanced analytics to investment decision-making — but they aren’t firing portfolio managers.
Instead, they’re building capabilities to help their human portfolio managers do a better job at investing, particularly as active managers continue to feel cost-cutting pressures in a shrinking industry, according to a new McKinsey study released today.
That means using data science to pinpoint and correct the mistakes investors are making and to focus portfolio managers on problems that only humans can solve. That’s in contrast to five years ago, when most traditional managers were doing very little with big data, leaving the field open to quantitative managers, according to the study, entitled “Advanced analytics in asset management: Beyond the buzz.”
“Data ultimately helps you focus your scarce investment resources — portfolio managers,” said Ju-Hon Kwek, a McKinsey partner and one of the authors of the paper, in an interview. “It’s less about whether to visit the factory floor or not, it’s about which one to visit. Data can help triage and manage human processing power and precision-target the human mind on specifics that matter.”
[II Deep Dive: UBS’s Big Data Lab Is No Longer Just for its Analysts]
Kwek emphasized that quants rely on algorithms, which they design, to make investing decisions and generate alpha. Fundamental managers use algorithms as well, but the algorithms feed insights to human beings, who ultimately make the investment decisions.
This isn’t exactly new, Kwek acknowledged. “It’s a bit of a fallacy that fundamental managers are purely qualitative,” he said. “It’s always been a business of fundamental insights supported by data.”
What’s different now is that the amount of available data has exploded — and traditional managers need to determine which quantitative sources to take in.
By leveraging academic research on human behavior, managers may be able to fix the common mistakes they make when investing using data science, according to McKinsey.
“The ability to stitch together a broad set of data sources about an individual or team’s trading history, communication patterns, psychometric attributes, and time-management practices allows firms to identify drivers of performance and behavioral root causes at a more granular and individualized level than previously,” according to the report. “Managers can then make operational improvements based on these insights.”
Kwek said that in addition to the managers using advanced analytics to enhance their human employees’ decision-making capabilities, another group is using these techniques to deliver more precise outcomes, such as with factor-based investing. In this case, managers are using data to zero in on and capture, say, the value or low-volatility factor.
Whether or not to use alternative data, natural language processing, and other techniques may ultimately come down to efficiency, McKinsey said.
As an example, Kwek said satellite data can now provide insights on retail sales in real-time, and natural language processing can look for valuable information in research and other reports at a speed and thoroughness that humans can’t replicate. “The question is how to use data to do things that are dull, dirty and dumb, and which high-cost portfolio managers shouldn’t be focused on,” said Kwek.
Successful asset managers are also using data science in sales and marketing. Although the industry is late to analytics in distribution, in part, because of the long bull market, managers are catching up.
For example, many have made strides in better segmenting potential clients.
“Our work with asset managers has shown that this type of behavioral-based segmentation of clients and subsequent adaptation of sales efforts can free up 15 percent or more of existing salesforce capacity and increase sales from priority client relationships by up to 30 percent,” the report stated.
Still, there are plenty of non-believers. Kwek gives a “qualified yes” when asked whether fundamental managers that don’t use advanced analytics to gain efficiencies have a future.
As with forecasting the weather, many experts rely on a long list of complex algorithms and data.
“Then there is the fisherman in Maine, who will say it will rain tomorrow. Sometimes, human intuition is very good at recognizing patterns,” Kwek said. “But the fisherman in Maine is not scalable. For those who can do data analysis well, it becomes a scalable process.”
