By Curt Engler, Head of Equities Trading, The Americas, JP Morgan Asset Management
Systemisation should be a gradual process with each stage of implementation tested for weakness until it becomes a solid foundation for further development.
New technologies have already had a profound impact on equity trading operations. Automation is well-embedded throughout the trade process and now systemisation is becoming more sophisticated, as trading desks leverage their quantitative skills and data analytics to further enhance order execution, strategy implementation and risk management.
However, the introduction of technology is neither indiscriminate nor hasty. Instead, a successful adoption of automated systems requires a deliberate and incremental approach. In practice, that means starting with the less glamorous stages in the trading cycle before moving to more complex applications.
At all levels, the technology is first used experimentally, tested and, if suitable, is then applied more extensively while constantly monitored and checked for any deficiencies.
The trading desk at JP Morgan Asset Management receives up to 8,000 orders a day from its portfolio managers, so automating the workflow was an early priority. By synchronising our order management system with our execution management system, then placing and monitoring small orders before extending usage more widely, we created a more streamlined and efficient process that lay the foundation for building automated trading strategies.
Again, these were introduced gradually. The intention was to extract biases that human traders are susceptible to, while also weighing the benefits – lower costs, greater trade execution efficiency – against the risks of automation – reduced flexibility.
We first made trial order placements in both lit and dark venues, and then gradually increased the trade sizes once we were sure the process functioned smoothly and the results were successful.
Now, between 70% and 80% of the notional amount of our trades is electronic and crossed in the market using sophisticated algorithms. The system accommodates a wide range of individual order size for portfolio managers with a variety of investment styles, such as momentum and reversal.
Broker selection is also systematised based on performance metrics for stock category and market conditions, and on their ability to facilitate orders within a particular investment style and increasingly complex trading strategies.
Although there are regional nuances, the structure has been adopted throughout our global operations. It is also being replicated for different asset classes, such as derivatives.
Importance of quant skills
In order to achieve this high degree of automation, trading desks have forged a much tighter, day-to-day working relationship with technology specialists than ever before.
Indeed, it would not have been possible without a considerable investment in qualified staff specialised in quantitative research and trading analytics.
Their expertise is critical throughout the trading cycle. They collect, disentangle and interpret data from a multitude of internal and external sources. For instance, they can identify tiers of liquidity for individual stocks at specific times and at particular trading venues in order to optimise trading strategies and fulfil our responsibility to achieve best execution – measured by transaction cost analysis.
For example, data can indicate disruptive or even manipulated market behaviour in an alternative trading system, which might be a signal to avoid trying to fill an order at that venue.
We’ve also introduced predictive analysis, based on data mining and statistical modelling, into our trading structure, and developed visualisation tools to increase our understanding of possible scenarios that follow a trade decision.
Of course, there will always be an important role for human agency within the trading process, especially to manage trades during unusual circumstances or for unorthodox transactions.
However, humans are clicking the trade execution button less and less, and the trend will continue. Machines are becoming more adept at completing even the most complex orders, ranging from crossing large blocks to finding a buyer for an illiquid odd-lot.
On the other hand, dealing desks will still hire staff – but they are more likely to have quantitative skills than streetwise agility.
At JP Morgan Asset Management, we will continue to leverage on the automation and systemisation that is already in place and has been rigorously tested and proven. More orders will be incorporated into the systemised process, which will be continually assessed for its performance and risk controls.
Automation needs to be disciplined, but not rigid. Markets can be volatile and so the systems also need to be nimble and flexible.
We’d love to hear your feedback on this article. Please click here