Growing role of automatic FX trading
OVER the last two years, we have witnessed a significant increase in demand to trade automatically in the FX market. Even if it is to follow the same trend as the equity and futures markets after Mifid – its high levels of liquidity and huge volumes of trade per day makes it a very attractive market for automated trading – this is a complex and very specific market to enter.
In the FX market, the need for speed is measured in milliseconds rather than microseconds. And although we are in the automated trading (HFT) space, the FX market is still far from entering the high frequency space because of its current structure and lack of best execution practices. Speed will be critical because the FX matching engine locations are very distant; the three main venues are New York, London and Tokyo.
BRIDGING THE GAP
In order to enter the HFT space, participants will need to optimise the following elements of their trading infrastructure: their network, connectivity and execution engine location. There are simple solutions to respond to those challenges, such as having a fibre optic network with a large bandwidth capacity, adopting low latency connectivity and using co-location and proximity hosting facilities. This kind of trading infrastructure was already in place for the other asset classes.
FURTHER CHALLENGES
Where FX has to face other challenges is with the highly fragmented bank and electronic communication network (ECN) execution venues, the lack of communication standards and the dispersed price discovery. Traders will have to capture market data from various and distant venues, all using different standards; therefore their trading systems need to be very open and adaptable. They also need to possess key features such as integrating real time data but also be capable of storing information to back test trading models. A platform needs to be scalable in order to back test multi-strategy models on different time scales. Ideally firms need to be able to test their models in different environments. For instance, statistical packages and engines need to be able to send orders to multiple venues in multiple protocols.
Developing and maintaining this kind of platform is very high in terms of cost but also resources. A lot of companies are still developing their platform in-house but now third parties can provide robust solutions and more firms will select this solution, allowing their internal valuable resources, quants and researchers to focus on their core value: building and developing alpha models.