FX Liquidity, “Follow The Sun”, and Price Engines Thoughts
First, lets clear up what “Follow the Sun” trading means.
The spot FX market is unique to any other market in the world, as trading is available 24-hours a day. Somewhere around the world, a financial center is open for business, and banks and other institutions exchange currencies, every hour of the day and night with generally only minor gaps on the weekend. Essentially foreign exchange markets follow the sun around the world, giving traders the flexibility of determining their trading day.
A quick read on Dummies “Liquidity and the Foreign Exchange Market” provides a view on liquidity. Of particular importance:
Forex market liquidity will vary throughout each trading day as global financial centers open and close in their respective time zones
Armed with this information, we can now turn to FCM360 “Foreign Exchange Hosting, Colocation & Connectivity” article that provide a good overview of the various FX venues and Colo’s.
Looking at the FCM360 network diagram, ignoring the fact that its FCM360, the choice of location for an FX pricing engine of a sell-side company is usually going to be either New York or London, or both. The “both” are often chosen to leverage “Follow the sun” from a liquidity perspective. What this means is that since Asia and European open before the US, during the course of a trading data, the ideal location of the Pricing Engine would be in London to reduce latency and sit closer to the liquidity centers. As the liquidity moves through the trading day, it maybe advantageous to swing the Pricing Engine to New York to gain a latency advantage. Obviously the swinging of a Pricing Engine from one locations to another needs to be well managed and leverage an appropriate software architecture, otherwise there could be costly implications.
Risk management could also take advantage of “Follow the Sun”, since the market data and hence risk calculations that leverage this data and the trade portfolio’s if located close to the source of the changing data, can themselves reduce the data movement over a LAN/WAN, and thereby speed up the overall calculation time.