Rickard-Carl Berglén , pp. 104. MAM/Sektionen för Management, 2010.
The focus of this thesis is to develop a model that aims to predict how electronic network latency
affects the performance, and in turn the profit and loss of Automatic Market Making making in
the Foreign Exchange Spot market. The output from this model will then be used in an
investment analysis to determine if the improved profit from upgrading a automatic trading
platform is large enough to give a solid return on investment. It´s widely known between players
in the Financial Services industry that network performance affects the profitability of automatic
trading algorithm, latency in networks will delay the view of market prices and order execution
and therefore make trading algorithms less predictable. To date there have not been any extensive
research focusing on the scale of the impact on profitability. A greater understanding of how
much network latency affects the performance of automatic trading strategies will give an
indication on how much it´s worth investing in surrounding IT systems and IT infrastructure in
order to maximize the profitability.
During this research a simplified model of Automatic FX Spot Market Making was developed.
Based on that model a trading scenario representing a typical Automatic Market Making
approach was built. The model and the scenario was then be implemented as a computer program
and simulations where run and relative profitability data collected.
The analysis of the data from the simulations clearly shows that profitability of an Automatic
Market Making Strategy drops drastically when network performance drops. The model shows
that there is a large difference in profitability between networks with no latency compared to a
network with some latency, with further latency the profitability shrinks in a slower rate.
The findings in this research clearly indicates that it is economically profitable to invest in high
performing electronic networks, which minimize latency as much as possible when running a
Automatic Market Making operation. However, before investing in a state of the art Automatic
Trading system a bank should make sure that the underlying financial models are correct.
This research has been done together with the head of FX Algoritmic Trading at UBS Investment
Bank and the results are of direct interest to both UBS and other participants in the Automatic