In Northern Trust Asset Management’s (NTAM) latest research, the firm explores how mapping company networks can help equity investors to spot opportunities and risks earlier.
Speaking to InvestorDaily, Jan Rohof, the firm’s director of quantitative strategies for APAC, emphasised that given the increasing diversification of modern businesses, this network-based insight can offer valuable context in how companies relate to each other.
“Companies don’t necessarily only compete in one sector or subsector or industry … They tend to touch very different parts of the economy,” Rohof said.
“This [mapping] allows us to get a much better understanding of companies as a result.”
As he explained, the firm’s “network aware” strategy uses natural language processing and large language modeling to analyse textual data such as competitor mentions in annual reports, aiming to achieve a risk-efficient outcome.
In turn, the firm’s investment decisions are now based on visually representable networks made from this primary data collection.
Looking at the history of quant strategies, Rohof explained that for a long time, most investors focused on traditional numerical data such as financial statements and balance sheets. However, with the advent of artificial intelligence and increased computing capabilities, the technology has ushered in a new era for the strategy.
For example, AXA Investment Managers has previously discussed how it is using quant analysis of textual data to streamline processes like onboarding patent data.
NTAM’s research discusses how these advancements have also increased the accessibility of alternative data analysis on a large scale.
“That is something that historically, without being able to venture into text-based information, we couldn’t incorporate in our investment decision making,” Rohof said.
He illustrated the firm’s research with the example of a car manufacturer’s supply chain.
Using the strategy, NTAM would create a network including every stage, from tyre and electronics suppliers to retail dealerships, accounting for the diversity of customer sales at each step, with appropriate weighting assigned to each relationship.
As Rohof explained, such a model can be used to pre-emptively identify a “lead-lag effect” between companies. For example, increased demand for cars would trickle down the supply chain to tyre manufacturers.
Rohof explained that these insights essentially give the firm an “informational edge” in positioning its portfolios, particularly when it comes to risk.
“If you flip it, and you see that demand is actually falling, or you see that a company is not being referenced a lot by competitors but their very direct peer is being mentioned much more by others, that also tells you something about the potential risk of a company,” Rohof added.
When combined with factors like deteriorating earnings and shifts in volatility, he argued that this additional information can be highly effective in projecting potential risk and return implications.
Can network-based insights offer a lasting edge?
Rohof noted that while these strategies have been highly effective in providing an informational advantage, he recognised the concern that such computer-based insights could become universally priced in as investors race to find alpha opportunities.
However, he asserted this outcome is unlikely because the value in these insights lies not just in the information itself, but in its integration into broader stock selections.
“We don’t see the risk of a lot of decay in this, because it’s really how you construct the signal on the back of the networks that really delivers the most value.
“It’s not the network itself, it’s how you utilise and contextualise that information and that network as part of your assessment,” he said.
For example, he explained how NTAM uses these networks in order to further inform the way it defines a momentum factor in a more consistent way, also taking into account other contextual factors.
For this reason, he also maintained that these insights should be viewed not as a replacement for fundamental strategies, but as a valuable complement to human judgement.
“We do see that fundamental strategies still have a place in portfolios,” Rohof said.
“In fact, combining quantitative and fundamental approaches can be a very effective way to diversify return sources and strengthen overall portfolio resilience.”
