Corvil » Corvil Insight

The SEC definition of latency transparency is, “the disclosure of information that would enable market participants to make informed decisions about their speed of access to an exchange or other trading center”. This statement in isolation of the SEC would bring few arguments from most traders whose strategy is sensitive to latency.  In fact these traders are already screaming at their IT staff and pleading with their market centers to provide better latency transparency and information. Why? The answer is cost.

The price to pay for zero latency is infinite spend. Even if one is willing to commit to infinite spend (which no one is), then your advantage may be short lived because of the “Law of Latency Relativity”. The law simply states:
“I do not really care how fast I am, I just need to be faster than you”

Therefore as soon as one or more competing players are also willing to invest the same amount, the relative speed advantage is reduced or removed completely. In practice the decision process for investment in low-latency trading infrastructures is more complex, but the fundamental issue of latency reduction to achieve trading performance advantage versus cost is an extremely important question for many organizations today. The key to answering this question is Latency Transparency.
In a world of finite ability to invest in low-latency technology combined with the Law Of Latency Relativity, information about speed is likely to become more important than speed itself.  This is why we say “speed is good, transparency better”.

Latency is not a constant. It changes, all the time. Most latency information published by exchanges and service providers does not always help. It is typically end-of-day reporting and often lacks critical pieces of information, e.g. load conditions and exact specification of how measurements were made. This type of approach to latency information publishing is more focused on demonstrating that the exchange or service provider is performing at low-latency as opposed to providing latency information in a manner that helps market participants use their services to achieve lower latency at the right cost. The latter information model is the true test of latency transparency and is the model used by CorvilClear where complete microsecond granular latency information is exchanged between trading parties continuously and in real-time.

If one cannot measure or cannot get access to latency information continuously and in real-time for speed of market price and speed of execution, then the only choice is to continue to spend much larger amounts of money on low-latency technology and services and pray that it pays-off.

Before you act - check the following:
When considering investment in low-latency trading technology and services:
1. What you see is not necessarily what you will get. Advertised latency performance metrics or benchmarks are often misleading and not representative of what is achieved in practice. This is largely due to the fact that your specific operating conditions are often different than those that existed at the time of the advertised latency performance. The second reason is that the measurement method and/or technology could be different, e.g. different measurement precision, accuracy, or period. The best way to tackle this is to make your own latency measurements for your specific operating conditions.
2. Make sure to identify where the real latency bottlenecks are. We often see people make large investments in certain parts of their trading loop, which fail to improve in a meaningful way the overall end-to-end latency. Therefore, no real improvement in fill rate is achieved. An example of this is co-location. Co-location is not a guarantee that trading performance improves. It is however a guarantee that propagation delay will be reduced. Therefore if propagation delay is a major component of overall latency, then co-location should provide a good return on investment. If however it is not, then do not be surprised if co-location disappoints. [ASIDE: The real latency benefit of co-location often extends from the fact that much higher capacity is achieved at co-lo centers compared to extranet access.]
3. Encourage the “help me, help you” model for latency transparency. Most providers of low-latency services use an approach we call “let me show you that I am fast”. Most consumers of low-latency services don’t like this approach and want a model we call “help me understand when you are both fast and not so fast, so I can select the best service and use it better”. We believe that a model of “help me, help you” adopted by both the providers and consumers of low-latency is ultimately the most helpful and efficient. This implies complete latency transparency between parties is required.

Acknowledgment:
This article was published in part on the TABB Forum as follows:
http://www.tabbforum.com/opinions/latency-transparency

Click here to download a copy of this article