Abstract:
|
In this study we investigate the high dimensional dynamic trading network among a set of stock brokers by assuming the existent of a low dimensional latent network. The latent network can be viewed as a trading network among communities of stock brokers and is discovered through a factor model for matrix time series. The analysis reveals interesting features of the trading network, trader community, liquidity supply and demand, and dark pools. The data set contains monthly trading volume of over 1000 stocks of various sizes and over 100 brokers in Australia.
|