B1954
Title: Modelling low latency
Authors: Vladimir Volkov - University of Tasmania (Australia) [presenting]
Yoann Potiron - Keio University (Japan)
Abstract: A novel approach to measuring low latency, defined as the time it takes to learn about an event and generate a response to this event, is proposed. The measure of low latency is obtained from an intensity model, which is an extension of the Hawkes model, allowing a memory kernel to be dependent on an additional unobservable stochastic variable characterised by low latency. Detailed information about cancellation orders and identification of traders, normally used in the literature, is not required in this case, which makes our approach more flexible in applications. Low latency estimates for the US and Canadian stock markets vary between 2 and 9 milliseconds from 2020 to 2021. The US firms are found to be more involved in relative latency competition, implying different risk appetites for firms with different latencies. Changes in low latency have a significant impact on the level of volatility in the US and Canada.
