Spotlight Series: Katie Lam
The distinction between algo trading and automation can make for an interesting discussion. Automated trading can be applied as pre-defined tools, and can omit knowledge of market advantages, whereas algo trading is more complex. It relies on high volumes of data, modelling and forecasting, and there is more accountability for execution and the impact on the order book.
Before embarking on an algo strategy, it’s crucial to understand the strategy or market inefficiency to be exploited – for example technological deficiencies or compliance requirements – and to know the dynamics in the market and amongst competitors.
But it’s technology and data challenges that are most often cited by market participants as the main impediments to attempting algo trading. Well-structured, robust and advanced IT infrastructure is required to support fast execution and minimal latency, in the order of milliseconds. This calls for systems using Python, Java, and R programming language with an emphasis on low latency. In order to train the data, trade your own algo as much as you can. Ideally, you should collaborate internally and maximise in-house efficiencies by training staff within the company. It’s an iterative cycle – develop modules within the team, train the team to develop models. Reliance on third party knowledge fails to develop the in-house culture necessary for digital transformation which in turn drives algo implementation. So it’s preferable to build systems internally than buy from the outside. It’s also key to quickly get the IT to the trading desk – where it’s going to be used and test. The desk moves faster than the IT team and this helps drive action and results.
Governance, regulatory and compliance considerations also need careful assessment and planning, especially with regard to risk strategy. Compliance rules from exchanges, clearing houses and regulators need to be supported and robust risk modelling and scenario testing conducted. This should encompass risk modelling under extreme conditions and detailed scenario analysis allowing for a severe tail in P&L. Risk parameters need to accommodate for special events and catastrophic losses.
In addition, modelling should account for alpha decay – how do you know you still have the advantage you thought you had? Is the model correct, and the market reacting to events, or is it alpha decay?
Monitoring is another hot topic garnering attention recently. Live monitoring of algo trading is an important part of the regulations, but real-time monitoring for 24/7 trading is not essential as algo only operates when a human trader is in the office for human oversight.
Effective implementation of algo trading requires careful planning and execution at the intersection of markets, technology, data and trading culture – and it will be fascinating to see how this landscape evolves.
Get in touch to arrange a meeting to discuss these themes in greater detail or learn more about our insights in technology and operations.
Get the latest news and stay up to date