The Last Look…
Posted by Colin Lambert. Last updated: June 28, 2022
I accept that this column often criticises, it is part and parcel of this job, but this week I come to praise CboeFX for a small but significant rise in its levels of transparency – a shift that I think could, or rather should, be replicated elsewhere in the industry both at platform and LP level.
As of this week, CboeFX is publishing both the percentage of overall volume traded anonymously, as well as the median response time on its non-firm liquidity. Both will help customers better understand how their flow is treated. Not every LP is in favour of anonymous trading, but for those clients to whom it is important, it is a good thing to be able to understand the liquidity with which they are likely to interact and the type of environment that exists on the platform (92% of flow on CboeFX was anonymous over the past 30 days).
Regular readers will be unsurprised to hear, however, that I am very happy with the second innovation in reporting! It means customers on the platform will be able to easily understand which, if any, of their LPs are either under-performing on the connectivity side or, worse, playing around with last look.
The current 30-day average response time as I write this is 10ms, on one day it went to 11, another it was at nine, but there is, in 10ms, a benchmark that people can easily understand and use to question the performance of anyone underperforming it.
The first and most obvious comment to make on this is it would be great if the other last look venues could publish something similar, I understand that many platforms, like CboeFX itself, provide private data on a batch basis to help consumers understand LP behaviour, but something like this pretty much allows them to do so on demand, rather than on a fixed weekly, monthly or quarterly basis.
I don’t think it is about demonstrating the actual speed of the trading process as such, rather it is about providing a level that customers can use to highlight outliers in technology performance. A platform could have an average response time of three or 20ms, it doesn’t really matter, what does though is how many LPs are a long way off that mark? It needn’t signify an LP playing games either, it could simply be their technology is a little slower, but at least by highlighting the issue, the median response time allows customers to take that into account if they are selecting LPs.
Well done to CboeFX for taking this small, but significant step, and the hope is that the other platforms in this space follow suit by publishing both percentage of anonymous volumes and median response times
This development actually plays to another theme that’s been buzzing around my head in recent weeks and months – has the debate on zero additional hold time has been too absolute? As you pick yourselves up off the floor at the thought of me defending last look in any fashion, I should clarify that I am firmly in favour of zero additional hold time, but it’s a bit of a blunt instrument and only goes half way to really describing what is going on.
Developments like that at CboeFX will help highlight any latency buffering going on of course, but I like the fact that it actually takes into account that there is a natural delay in trading. This is a physical reality – there just isn’t the ability to, for example, trade instantly between New York and Singapore.
What I would like to see in LP disclosures – apart from the commitment to zero additional hold time, and the clock is ticking on that by the way – is exactly what CboeFX is providing here – a median round trip time. This would take into account the challenges of quoting someone in Singapore from New York as well as local trading and allow a performance benchmark to be established for clients.
I am not sure how it happened, but the debate on additional hold time has been translated in too many quarters into the suggestion that there is no such thing as latency – there is, and to my mind the average should be disclosed to help clients understand how their LP is handing their flow. That way, we no longer have debates about hold times, we discuss any deviation from the median – which is any real problem would lie. On occasion there are going to be deviations and they will be explainable, and that is all most people really want – an understanding of when things go innocently wrong or a signal that someone is playing games. In an ideal world, an LPs disclosures will state the average response time as well as their commitment to zero additional hold time.
So well done to CboeFX for taking this small, but significant step, and the hope is that the other platforms in this space follow suit by publishing both percentage of anonymous volumes and median response times. Not only that, but I would hope this practice could be adopted by LPs the world over to help explain to their clients how they are handing their flow.
There’s nothing wrong with being (relatively) slow, but there is in deliberately being so, the more we understand about our trading environment the better.