By: Joe Schifano, Global Head of Regulatory Affairs at Eventus Systems
On September 1st, Eventus Systems and VoxSmart submitted a joint comment letter regarding FINRA’s June 2020 report, “Artificial Intelligence in the Securities Industry.” The broader acceptance and use of artificial intelligence (“AI”) technologies continues to advance across the financial services industry and has become much more accessible to broker-dealers through both internal development and experienced service providers.
Many firms already use elements of machine learning (“ML”) with their front-office applications and surveillance platforms. As firms implement new technologies to help scale and improve surveillance, the increased use of ML presents operational and regulatory challenges. Here, we will focus on data governance and FINRA Rule 2010.
Evolving technology typically enables greater efficiencies and benefits that help enable individuals and companies achieve their goals. However, such evolving technology inevitably raises legal and regulatory questions for consideration. As AI techniques evolve in front-office applications, new issues relating to data governance test the boundaries and understanding of existing rules and guidance. Continuing regulatory guidance is critical so that broker-dealers can further engage these new technologies with a more confident and proactive approach.
FINRA Rule 2010 – Data Governance
FINRA Rule 2010 states, “A member, in the conduct of its business, shall observe high standards of commercial honor and just and equitable principles of trade.” A succinct and principle-based rule, it is cited by FINRA in a wide range of its Regulatory Notices (e.g., 15-09 on Effective Supervision, 17-22 on Disruptive Quoting). With its roots in the Securities Exchange Act of 1934, the rule demands a high level of ethical and professional behavior from FINRA members. Its reference in FINRA’s June 2020 report means that the data that feeds front-office and surveillance platforms should be complete, accurate and fit-for-purpose. With the explosion of data and its uses, FINRA recognizes the need for additional dialogue and guidance on data governance generally, and especially for applications using AI techniques. However, citing Rule 2010 puts broker-dealers on notice that while best practices evolve, broker-dealers have an obligation to ensure their data governance protocols meet the spirit and standard of this often-cited rule.
Data bias and error rates are issues to consider when building a surveillance platform. At Eventus, we use ML to make sense of large volumes of data, enabling firms to cut through the clutter and ensure the most important alerts are flagged to the top. For trade surveillance, data originates from multiple trading systems and in differing data protocols. Analyzing sub-data elements across many products and systems requires advanced analytics to extract and interpret relevant data to drive intelligence and pattern analysis. While analysis drives key decision-making and surveillance tools, such analysis is subject to an error rate, which itself is difficult to determine. This is the key benefit of defined compensating controls for business applications (e.g., trading algorithms). Eventus employs a number of techniques to mitigate potential data bias through supervisory reviews, both manual and automated. In the end, the models are only as good as the data, and surveillance errors can naturally occur. Front-office controls and a mature supervisory process may compensate for such surveillance errors.
Our Recommendations to FINRA
Our comment letter suggests that the broker-dealer community would benefit from continued practical guidance on data governance and transparency on FINRA’s expectations. Both guidance and transparency will enable firms to move forward progressively and converge on common standards. Our suggestions include:
- Defining best practices for data governance (e.g., standards for back testing, data remediation, process failures, data protocols, compensating controls and expected governance requirements for applications that use ML-models)
- Sharing FINRA’s experiences or relevant resources that solve for testing and data bias
- Partnering with AI-development companies to promote and socialize best practices
Eventus and VoxSmart strongly support FINRA’s initiatives concerning the adoption of these new technologies into the regulatory framework governing broker-dealers.
We believe that a rapid, controlled and uniform adoption of AI for trade and communications surveillance will enable supervisors and compliance officers to manage risk in a highly complex, interconnected and evolving environment. To read about how FINRA Rule 3170 applies to the use of AI technology in communications surveillance, please visit the VoxSmart website, HERE.
If you have any questions about the joint comment letter or the application of AI in the securities industry, please don’t hesitate to reach out to us at: firstname.lastname@example.org.
Eventus Systems is a leading global provider of multi-asset class trade surveillance and market risk solutions. Its powerful, award-winning Validus platform is easy to deploy, customize and operate across equities, options, futures, foreign exchange (FX), fixed income and digital asset markets. Validus is proven in the most complex, high-volume and real-time environments of tier-1 banks, brokerages, futures commission merchants (FCMs), clearing firms, trading firms, market centers, buy-side institutions and corporates. The company’s rapidly growing client base of more than 60 firms relies on Validus and Eventus’ responsive support and product development teams to overcome its most pressing regulatory challenges. For more, visit www.eventussystems.com.