Kevin Houstoun shares his thoughts on impactful trends in trading technology during an interview at the recent TradeTech Conference in Paris.
In addition to his role as Rapid Addition’s Chairman, Kevin is a Non-Executive Director at FIX Protocol Ltd., a visiting researcher at the London School of Economics Systemic Risk Centre, and holds a number of other non-exec and advisory positions.
What new initiatives will impact trading technology over the next few years?
[From a FIX Protocol perspective] I think some of the things we’re working on from the standards perspective are very relevant here.
We’re looking at continuing the work we started with the [FIX] repository, and are now calling Orchestration, to be able to code generate a lot of the ‘plumbing’ type technology. And as we start to make that machine readable, we are starting to see electronic rules of engagement, as opposed to the large, printed documents that we’ve historically taken as the ‘how we link’, say, an exchange to a buy-side, or how we link a buy-side to a sell-side order management system such as Quod Financial.
How we automate that whole process, and automating that whole connectivity piece, is going to be a big trend over the next few years.
What benefits will this bring to the industry?
Things like automating the connectivity, as I just mentioned.
If you think about it, one of the factors you’ve got is which venues you can connect to, how you can connect to them, how you can connect to brokers?
That gives you more choice, and of course sometimes that will give you a better way of getting a better price from somewhere, or receiving more information, or being able to say ‘ok these 2 instruments seem correlated’ so we can use that as a signal.
So, being able to automate that means you can access more liquidity, more different venues and more different sources of liquidity at a lower cost, and that of course eventually overcomes a barrier of ‘okay it’s just too expensive and too timely to have complete coverage’.
Where does AI and Machine Learning fit into the future of trading workflow?
If you think about us as human beings, we can perceive data and we can spot patterns. But we can’t do that in vast data sets.
As we automate technology to recognise those patterns across much bigger data sets than we’re able to perceive, we spot new patterns. Those new patterns will allow us to make better trading decisions.
We won’t necessarily even understand why the machines are picking up those signals, and one of the big things we will need to able to do (and it’s an area that is currently undergoing a lot of academic research) is explain the decisions that neural networks, and the like, make.
How is Rapid Addition leveraging trends in automation and machine learning?
We work very actively around automation of lots of things.
We have automation of our entire build for example.
We provide infrastructure software, and it is important to our clients that this infrastructure software performs consistently; that we don’t do a new release that causes a sudden spike in latency, for example.
So, one of the areas we’ve automated is that we have the entire platform tested every night, not just for functional but also for non-functional aspects, such as latency as so on.
Another area we’re actively looking at is using machine learning to recognise common patterns in the vast code database that we have, and say ‘okay, these patterns can be performed like this’, and can we actually get to the point where we are able to start using machine learning to write more of the basic code.
If you look at any coding toolsets used to produce things like apps for mobile phones, one thing you’re seeing is more and more of the boilerplate code being machine generated.
That machine generation started off being template driven but with artificial intelligence and machine learning it is starting to be more intelligently generated.