Algorithmic trading, including high frequency trading, was not expressly captured under MiFID I; however, as this type of trading has rapidly developed within the financial services industry so has the potential for attendant systemic risks. MiFID II has introduced specific rules relating to this type of electronic trading activity to help mitigate against ‘flash crashes’ and also combat potentially abusive behaviour.
Algorithmic trading has been defined widely in MiFID II as “a computer algorithm… with limited or no human intervention”. The Level 1 and Level 2 documentation set out a number of obligations with respect to the conduct of algorithmic trading including;
- Ensuring Staff connected with Algorithmic Trading are competent for the roles they will be performing including IT, Compliance, Risk and Traders, with access to systems being restricted to those who require it;
- Systems and controls are in place to ensure that firms effectively build, monitor and test algorithms before deployment, both internally and also with trading venues in UAT environments;
- Ensuring that there is an effective BCP that is in operation and that this plan actively considers the algorithm, together with a dedicated ‘Kill switch’ which is available for immediate use if required;
- Clear and effective procedures in place for maintaining the infrastructure related to algorithmic trading, as well as being able to demonstrate governance around the operation of the algorithm;
- Prescriptive record keeping of each order generated by the algorithm; and
- Effective monitoring of market abuse or attempted market abuse, or other forms of disorderly trading occurring from the algorithm, both real time and post-trade.
The associated Regulatory Technical Standards run to many pages, and are aimed at ensuring that systems and controls are robust and effective. Firms that operate, design or develop algorithms would be wise to invest time to carefully consider the prescriptive obligations now.