Fundamental Networks

How the quantitative mapping of corporate ‘fundamental networks’ can give a more refined framework for understanding companies.

It’s no secret that today’s companies exist in an increasingly complex, interconnected ‘ecosystem’. A single organization has multiple relationships across geographical regions: other companies along the supply chain, competitors, and partners. This evolving web of connections poses challenges for traditional investors who have finite capacity and speed to process information, resulting in potentially exploitable mispricing opportunities. These types of relationships are intuitive, and we believe that ‘fundamental networks’ in markets can help uncover opportunities.

At Man Numeric, our team has developed systematic techniques aiming to extract under-utilized, stock-specific information using fundamental relationship data. Figure 1 gives an example of a fundamental network around a particular company, Apple Inc, where circles (or ‘nodes’) represent the companies connected to each other and their market cap. This is just one way of illustrating such a network – others include more granular analysis of individual business segment revenues or market cap.

Figure 1: Example of fundamental networks – Apple Inc

Source: Man Numeric. For illustrative purposes only. The content of this material is not intended to constitute, and should not be construed as, a recommendation or solicitation to transact in the securities of the companies named. The organisations and/or financial instruments mentioned are for reference purposes only. This information is solely used to demonstrate Man Numeric’s internal research capabilities.

How do we use this type of information to understand a company’s ecosystem? There are three key considerations in using network analysis: record network information, information propagation and node centrality. First, recording network information involves describing precisely how companies interact with each other – for example, the direction of information flow between them, or the properties of each node in a network. Second, information propagation is about the way we observe information from one company impacting another company. For example, two nodes in a network are strongly connected if they are linked by multiple paths. Figure 2 gives an example of both these dynamics, illustrating a simplified undirected global network, comprised of competitors, customers and partners. At top is a representation of a simplified network of companies (1 to 8), connected in various ways. The adjacency matrix in the left shows whether the companies are directly connected (‘1’ indicates they are, ‘0’ indicates they are not), and the matrix on the right plots the number of two-step routes between companies. We have highlighted the links between companies 3 and 7 on each matrix – not connected directly (hence the 0 in the first matrix) but accessible via three different two-walk paths (via nodes 2, 8 or 4, hence the 3 in the second matrix).

Figure 2: Information propagation in a simplified global network

Source: Man Numeric. For illustrative purposes only.

We believe these first two steps are important for quantifying the connectivity of companies in a network. But the third area of focus is the importance of individual nodes – which is not always the same as the number of links it has to others. Indeed, equal weighting of nodes in a network may fail to capture the real dynamics at play between companies, where a node is ‘central’ if it has many connections to others, and where its status can depend on the status of its neighbours. There are multiple ways of quantifying the importance of a company in a network, and the choice between them depends on the specific applications and types of network. For investors, the key question here is about whether they generally look to take positions in more or less ‘central’ companies – which again depends on the investment strategy to which this analysis is being applied.

Ultimately, fundamental networks are built on intuitive observations about the way companies interact. Their basis is nothing new, but we believe that this systematic approach to quantifying relationships across markets can help investors understand the equity market universe using a more consistent framework. As interconnectivity between companies continues to increase, advanced network data analysis can be used to complement existing quantitative equity research, and we believe that if used intelligently, it can potentially provide further opportunities to add value.