An interesting comment in today’s FX Week, touches on the real-world challenges banks face when designing a new single bank platform.

Banks want to deliver a really great trading application to clients. One that combines the very best the bank has to offer in terms of timely pre-trade decision support tools such as research, streaming video, analytics, dealer chat, all wrapped intuitively around fast, executable dealable prices. What the bank definitely doesn’t want is for its shiny new GUI, crammed full of pre-trade ‘stuff’ to in effect swamp the client’s PC, making trade execution sluggish, or impossible for clients in poor band width locations, resulting in the bank having to disable one-click dealing, due to unacceptable latency.

At Caplin recently we’ve been working on smarter client GUIs that can self regulate the amount of data they try to consume, based on using feedback loop algorithms, that take into account factors such as the latency of prices as received by the client, and changes in CPU usage of the PC, and the bank can set triggers and business rules around what to do when the measured client latency (or CPU) rises above certain levels. The GUI then self regulates by elegantly and deterministically degrading the non essential (non-trading) parts of the trading application, so that executable streaming rates continue to be dealable, and are delivered in a timely manner.

The bank too should incorporate within its pricing engine and streaming architecture similar measured latency feedback algorithms (called adaptive throttling) to modify, or throttle the update frequency of prices being sent to the trading application. In fast trending markets, the price latency has an even greater impact, and therefore any adaptive throttling algorithms need to take into account not only changes in latency, or the rate of change in latency, but also the rate of change in the underlying prices, the more volatile the rates, the less the bank can tolerate any latency in price delivery. Such market conditions triggering widening of spreads, or changes from dealable to indicative rates being streamed to certain clients.

In this way, banks can avoid choking their clients, which is never a good idea!