10 Correlations, Hierarchies, Networks and Clustering
257
10.5.4 Based on Key People
• In [111], authors build a co-occurrence network of people mentioned in 21,578
Reuters news articles mostly focused on economics published in 1987: People
are represented as vertices and two persons are connected if they co-occur in
the same article. They observed that this network has small-world features with
power-law degree distribution. The network is disconnected and the component
size distribution has power-law characteristics. They compare the importance of
these people back in 1987 in the news co-occurrence network to the importance
they have as of 2007 proxied by their Wikipedia article. They find medium level
Spearman’s rank correlations between these two measures.
10.5.5 Based on Investors
• Traders and investors are socially connected and have access to comparable sources
of information. Reference [29] investigates social connections and information
linkages, and demonstrates how they can predict patterns of trade correlation:
trades generated by “neighbor” traders are positively correlated and trades generated by “distant” traders are negatively correlated.
• Using a dataset of all trades on the Istanbul Stock Exchange in 2005, [112] identifies
traders with similar trading behavior as linked in an empirical investor network.
They find that central investors earn higher returns and trade earlier than peripheral
investors with respect to information events (e.g. earnings), consistent with the
view that information diffusion among the investor population influences trading
behavior and returns.
• In [7], authors study investor clusters in the first two years after the IPO filing in
the Helsinki Stock Exchange by using a statistically validated network method to
infer investor links based on the co-occurrences of investors’ trade timing for 69
IPO stocks. They find that the clusters of investors based on their trading are stable
between new and mature stocks. Authors identify a highly persistent institutional
investor cluster, which they consider as evidence about institutional herding.
• Reference [8] investigate trading decisions for Finnish investors using multilayer
networks. They find that households in the capital have high centrality in investor
networks, which, under the theory of information channels in investor networks,
suggests that they are well-informed investors.
• Using a Kinetic Ising model, [25] reconstructs a lead-lag network of traders client
of a major dealer in the Foreign Exchange market. They identify leading players
in this market who are typically leading the order flow on the 5 minutes timescale. Based on this lead-lag network, authors also define a herding measure based
on the observe and inferred traders’ opinions. They find a causal link between
herding and the liquidity in the inter-dealer market used by dealers to rebalance
their inventories.
257
10.5.4 Based on Key People
• In [111], authors build a co-occurrence network of people mentioned in 21,578
Reuters news articles mostly focused on economics published in 1987: People
are represented as vertices and two persons are connected if they co-occur in
the same article. They observed that this network has small-world features with
power-law degree distribution. The network is disconnected and the component
size distribution has power-law characteristics. They compare the importance of
these people back in 1987 in the news co-occurrence network to the importance
they have as of 2007 proxied by their Wikipedia article. They find medium level
Spearman’s rank correlations between these two measures.
10.5.5 Based on Investors
• Traders and investors are socially connected and have access to comparable sources
of information. Reference [29] investigates social connections and information
linkages, and demonstrates how they can predict patterns of trade correlation:
trades generated by “neighbor” traders are positively correlated and trades generated by “distant” traders are negatively correlated.
• Using a dataset of all trades on the Istanbul Stock Exchange in 2005, [112] identifies
traders with similar trading behavior as linked in an empirical investor network.
They find that central investors earn higher returns and trade earlier than peripheral
investors with respect to information events (e.g. earnings), consistent with the
view that information diffusion among the investor population influences trading
behavior and returns.
• In [7], authors study investor clusters in the first two years after the IPO filing in
the Helsinki Stock Exchange by using a statistically validated network method to
infer investor links based on the co-occurrences of investors’ trade timing for 69
IPO stocks. They find that the clusters of investors based on their trading are stable
between new and mature stocks. Authors identify a highly persistent institutional
investor cluster, which they consider as evidence about institutional herding.
• Reference [8] investigate trading decisions for Finnish investors using multilayer
networks. They find that households in the capital have high centrality in investor
networks, which, under the theory of information channels in investor networks,
suggests that they are well-informed investors.
• Using a Kinetic Ising model, [25] reconstructs a lead-lag network of traders client
of a major dealer in the Foreign Exchange market. They identify leading players
in this market who are typically leading the order flow on the 5 minutes timescale. Based on this lead-lag network, authors also define a herding measure based
on the observe and inferred traders’ opinions. They find a causal link between
herding and the liquidity in the inter-dealer market used by dealers to rebalance
their inventories.