Measuring Up: FIX Inter-Party Latency Update
Raymond Russell, Julian Ragless, Richard Leung | Corvil, Hong Kong Exchanges and Clearing, Ltd | December 15, 2011
Measuring Up: FIX Inter-Party Latency Update
Julian Ragless and Richard Leung, Hong Kong Exchanges and Clearing Ltd, discuss the value of latency measurement to an exchange and the ways they are reaching out to latency sensitive traders.
Given the multifaceted nature of the trade lifecycle - some latency sensitive, others not - how important is latency to delivering value as a trading platform as well as across the exchange and clearing business lines?
As a general principle, low latency is an important objective for any exchange. Our primary role as a secondary market operator is to provide an efficient market and the lower the latency, the sooner trading information can be priced in and disseminated to the market. The key term is ‘delivering value.’ To some people low latency is the primary value driver, while for other firms it is not. In recent years, many exchanges, particularly those in competitive, fragmented markets, have deliberately and aggressively pursued lower latency to attract High Frequency Trading (HFT). To some this is a good thing, in particular those providing HFT services, yet others have reservations, believing HFT does not improve their trading and may add social cost to the wider market through increased market data rates and bandwidth costs.
Lower latency is not always better. The peak message rate in the US markets is close to 7 million messages per second, which can be difficult for smaller players to absorb and process. Already, some in the US and Europe are making noises about whether this is improving the efficiency of the market and whether smaller players are being driven out of the market by escalating IT expenditures. Recent proposals for a transaction tax and a minimum time for orders to rest on the order book show that momentum is shifting away from low latency. Furthermore, there is no direct correlation between the number of trades and overall market turnover. Low latency trading increases the number of trades, but they may be of a smaller average order size.
What value does latency measurement give to an exchange and its members?
Latency measurement systems are extremely important tools for an exchange because we must accurately measure the latency in our systems and be able to disseminate it to the market on a real-time basis. Previously, time synchronization protocols like NTP (Network Time Protocol) only gave us resolution up to a millisecond, which was of limited use, whereas now we can measure down to nanosecond intervals using new protocols such as PTP (Precision Time Protocol). To an exchange, the ability to have an objective measurement of latency allows us to determine where there is congestion in the system, allowing us to take remedial action and helping us in the system upgrade decision. As an exchange, there is also a need to measure latency deviation to ensure that we provide a consistent level of latency. It is one thing to be fast, but what the market wants is a consistent, as well as low, level of latency.
Exchange participants benefit from a real-time latency measurement systems because an exchange trading platform is a dynamic system. More and more, end users utilise the realtime latency data as an extra piece of information to determine their trading strategy. In markets with multiple venues, the latency may determine where orders are sent. In markets with a single exchange, real-time latency data still informs traders about when to trade or which model to use. Both for exchanges and participants, real-time latency measurement is as important as a low-latency platform.
Tags: AMS 3.8, architecture, average order size, bandwidth costs, competitive differentiator, Corvil, data centre, Data Recovery, DMA, DR, Europe, exchange gateway, failover, fill rates, FIX Inter-Party Latency, FIX Protocol, FIXIPL, HFT, high frequency trading, HKEx, Hong Kong Exchanges and Clearing, infrastructure, Inter-Party Latency Working Group, interoperability, IT expenditures, Julian Ragless, latency deviation, latency monitoring, load balance, low-latency, market data, Matching Engine, millisecond, multicast, Network Time Protocol, NTP, peak message rate, Precision Time Protocol, PTP, Raymond Russell, real-time, Richard Leung, social cost, tick-by-tick, time stamps, trading desks, transaction tax, transparency, ultra low latency, US