The guide describes the governance, risk controls, and development and testing process that SGX members should have in place when using algorithms in trading.
SGX RegCo (Singapore Exchange Regulation) has published a new regulatory guide offering best practices to member firms that use algorithms to trade Singapore markets.
The guide follows a series of supervisory visits SGX RegCo has conducted over the past year to understand the behaviour, governance, and controls employed by member firms to prevent, detect, and mitigate market disruptions and system anomalies in their use of algorithmic trading.
Governance and oversight
The guide emphasises that proper governance and oversight over algorithmic trading frameworks should be deployed and maintained, comprising:
- policies and procedures around all aspects of algorithmic trading, including the development, testing, and deployment of algorithms, as well as minimum standards on the monitoring, controls, and review process of the algorithmic strategies utilised
- management committees comprising individuals of appropriate seniority and experience to provide oversight of the algorithmic trading activities, and to be responsible for the algorithmic trading framework, including to ensure that any new algorithmic strategies do not cause market disorderliness or disruption
- clearly-defined escalation channels from the frontline or working-level staff to their supervisors and the appropriate management committees to ensure that any potential issues with algorithmic trading which may arise can be promptly escalated to persons with the relevant authority
- notifications to SGX, in the event that member firms encounter any malfunction in an algorithm which may cause market disorderliness or disruption, so that SGX may help to minimise any the impact
- proper records and audit trails – including inventories of algorithmic trading strategies and controls, management committee meeting records, supporting documentation related to the algorithm development and testing, periodic reviews, and changes made to records
- adequate training for relevant staff and relevant clients to ensure that they are equipped with the technical knowledge and expertise needed to maintain orderly trading and understand how risk controls operate and how to handle potential issues that may arise
Risk controls
The regulatory guide also recommends member firms put in place robust risk controls – such as pre-trade risk controls and kill switches – to ensure that any potentially undesirable market disruptions are captured in a timely manner, and that orders can be rejected and algorithms can be suspended, if needed.
Such risk controls may include price controls, volume controls, client credit controls, burst controls, restricted orders, and kill switches. The guidance offers good practices for implementing such controls based on SGX RegCo’s supervisory visits.
The guide also highlights the importance of post-trade controls for addressing any trading activity or behaviour not caught in pre-trade controls. These controls include trade surveillance, regular system health checks, and periodic independent audit reviews on algorithmic trading.
The alerts generated from pre-trade controls should be reviewed on a “near-real time basis”, and post-trade controls “within a reasonable time frame”, to ensure any potential market impact resulting from algorithmic trading behaviour can be addressed on a timely basis, the guidance says.
“Additionally, it is recommended that the framework, policies, algorithms, and control parameters surrounding algorithmic trading should be reviewed at least annually, and at appropriate levels of seniority, to ensure that they remain effective and are up-to-date with current market developments.”
Development and testing
The regulatory guidance emphasises the importance of having a strong framework in place for the algorithm development and review process. Member firms are reminded to document the entire process and all of its stages and ensure adequate oversight by senior and experienced personnel.
The process should include robust controls to ensure code quality and security, independent testing prior to release, management review and approval, and ongoing monitoring to detect dysfunctional algorithms.
“There is a feedback loop between the Development and Testing stages such that when not all test criteria are cleared in the Testing stage, algorithms will loop back to the Development stage for further development work until there is sufficient assurance that the test criteria are cleared and the algorithms are working as intended,” the guidance says.
When using algorithms developed by third-parties – whether a vendor or a customer – member firms should subject all orders and trades executed by these algorithms to the same level of scrutiny and controls as if they were developed in-house.
This includes appropriate due diligence to ensure firms understand how third-party algorithms are developed, how they function, and what risks they might present.
Member firms should ensure they understand the behaviour of algorithms and that they work effectively in stressed scenarios, which entails rigorous testing.
Suggested stress scenarios include extreme market conditions (e.g. extreme volatility, liquidity stress, natural disasters), capacity testing (e.g. volume spikes, vendor connectivity issues), multiple algorithms working in tandem, and external connectivity issues (e.g. latency disruptions, order rejections).
Improving robustness
SGX members are expected to use the guide to improve the robustness of their controls, which in turn will enhance confidence in Singapore’s markets.
The regulatory guide, available here, also offers three case studies for illustrative purposes
