“…a significant amount of the External Fund’s capital was allocated to a Semi-Systematic Capital Unit called RMT, which was a semi-automated computer trading system…RMT did not perform in the same way as the Portfolio Managers, and at times its performance fell below theirs”
Paragraphs 2.7-2.8, FCA Decision Notice to BlueCrest Capital Management (UK) LLP (“BCMUK”), 4th November 2021
During the Christmas period I was hooked on Flash Crash: A Trading Savant, a Global Manhunt and the Most Mysterious Market Crash in History by Liam Vaughan (2020, William Collins, ISBN: 0008270392). Vaughan’s non-fiction thriller provides a racy account of how one man, Navinder Singh Sarao, devised a trading strategy from his bedroom that the US authorities found to be a significant contributor to the 2010 Flash Crash which saw circa $1 trillion wiped off the value of the flagship US stock indices in a matter of minutes (Verity and Lawrie, BBC Business, 2020). Imagine if a vox pop were performed in the City to solicit views on the compliance challenges posed by trading algorithms. There is a significant chance that layering and spoofing type behaviours that came to be associated with the “Hound of Hounslow” would feature prominently in responses. However, it would be a mistake to merely view algo conduct risk through the prism of high frequency trading and black boxes.
By now, many readers will be familiar with the facts of the BCMUK case. If you need a refresher, look at Impact analysis: FCA intends to fine BlueCrest Capital Management £40,806,700 for conflicts of interest failings by Mike Cowan, published on the Thomson Reuters Regulatory Intelligence platform on 5th January 2022. The purpose of this piece is to examine what, if anything, this case teaches us about the identification and mitigation of conduct risk associated with the operation of trading algorithms.
Recap – the FCA’s expectations
In its paper entitled Algorithmic Trading Compliance in Wholesale Markets (February 2018, hereinafter the “FCA AT Guide”) the FCA provided examples of good and poor practice in the context of the governance of trading algorithms. In the paper, the FCA stated that one of its key objectives is:
“To ensure firms appropriately consider the potential impact of their algorithmic trading on market integrity, monitor for potential conduct issues and reduce market abuse risks”.
First, it is important to stress that the FCA’s findings in the BCMUK case are provisional and do not involve any suggestion of market abuse or damage to the integrity of the market. The FCA’s concerns stem from perceived failures to effectively manage a conflict of interest. The FCA “considers BCMUK’s conduct was reckless, rather than deliberate”.
Second, at the time of writing, BCMUK is yet to respond to the findings in the FCA’s Decision Notice.
One could interpret the FCA’s reference to “conduct issues” in the narrow sense, i.e. in terms of the deployment of potentially abusive strategies such as front running or spoofing algorithms. Alternatively, one could take a broader view such as that espoused by KPMG Partner Lucas Ocelewicz in his article Stamping out conduct risk in algorithmic trading (2017). Ocelewicz argues that algorithmic trading systems can undermine any of the FCA’s three strategic objectives “…either due to mistakes and errors in implementation or intentional design features”.
For the purposes of this article, I have adopted Ocelewicz’s broader interpretation of algorithmic conduct risk.
Insights for firms deploying trading algorithms
Considering the above, certain aspects of the BCMUK case offer valuable insights for both sell and buy side firms deploying trading algorithms, whether of the execution enhancement or investment decision making variety. If you hold the certified algorithmic trading function under the Senior Mangers and Certification Regime (the “SMCR”), take note!
In the BCMUK Decision Notice, the FCA states that funds were allocated to “Semi-Systematic Capital Units” that “operated through a combination of algorithmic processes and manual (i.e. human) review and intervention by staff at several Sub-Investment Managers (including BCMUK)”. Referred to as “Rates Management Trading” or “RMT”, the FCA states that these semi-systematic processes “attempted to identify an optimal set of profitable, cost-effective trades which would be executed based upon, but independently of, the trading activity” of portfolio managers. The FCA then goes on to describe some challenges posed by the RMT’s operations and matters arising therefrom:
- “significant slippage” between targeted and actual performance;
- “deficiencies in the data” that was fed into RMT;
- the continued allocation of “significant capital to RMT” on behalf of the BlueCrest group’s External Fund (where external investors’ money was received), in spite of “significant losses” and the subsequent reduction, and eventual cessation, of the Internal Fund’s (used to invest funds contributed by partners and employees of the BlueCrest group) exposure to RMT; and
- a lack of “proactive” disclosure of the RMT operation to investors in the External Fund, because of an emphasis on it being “proprietary” to the BlueCrest group.
Now, consider that in FCA explains: Conduct Rules training (2020), the FCA stressed that role specific training should:
- “be interactive and based on realistic and challenging scenarios” and
- “bring to life real questions that relate to people’s jobs”.
The FCA’s AT Guide also makes several references to the importance of suitable training for persons involved in the deployment and supervision of trading algorithms.
Again, the findings in the FCA’s Decision Notice relating to BCMUK are provisional and do not make any reference to Conduct Rule breaches or “criticism of any person other than BCMUK”. Indeed, the “Relevant Period” referred to in the Decision Notice (1st October 2011 to 31st December 2015) predate the entry into force of the SMCR. Nevertheless, some of the matters raised in the decision notice could be used by firms to design effective Conduct Rules and FCA AT Guide training for professionals involved in the design and deployment of trading algorithms. The table below shows how each of the abovementioned elements of the BCMUK case could be used to illustrate the practical application of specific Conduct Rules and expectations in the FCA AT Guide.
|Challenge / matter
|Individual Conduct Rule(s)
|FCA AT Guide
|Identification of potential performance issues or data deficiencies
|Rule 2: You must act with skill, care and due diligence
|Section 3: Development and testing
|Continued deployment after deficiencies detected
|Rule 4: You must pay due regard to the interests of customers and treat them fairly
|Section 6: Market conduct
Algo conduct risk is multi-faceted. Potential instances of market abuse grab the headlines, but firms’ training needs to equip staff with the ability to identify and mitigate less obvious forms of conduct risk that could arise because of the deployment of trading algorithms.
Originally published by Thomson Reuters © Thomson Reuters
2022. Cowan, M. Impact Analysis: FCA intends to fine BlueCrest Capital Management £40,806,700 for conflicts of interest failings, Thomson Reuters. Available to subscribers of Thomson Reuters Regulatory Intelligence.
2021. Decision Notice: BlueCrest Capital Management (UK) LLP, Financial Conduct Authority. Available at: Decision Notice 2021: BlueCrest Capital Management (UK) LLP (fca.org.uk) (last accessed 13th January 2022).
2020. Vaughan, L. Flash Crash: a Global Manhunt and the Most Mysterious Market Crash in History. William Collins, ISBN: 0008270392.
2020. Verity, A. & Lawrie, E. Hound of Hounslow: Who is Navinder Sarao, the ‘flash crash trader’? BBC Business, 28th January 2020. Available at: Hound of Hounslow: Who is Navinder Sarao, the ‘flash crash trader’? – BBC News (last accessed 13th January 2022).
2018. Algorithmic Trading Compliance in Wholesale Markets, Financial Conduct Authority. Available at: https://www.fca.org.uk/publication/multi-firm-reviews/algorithmic-trading-compliance-wholesale-markets.pdf (last accessed 13th January 2022).
2017. Ocelewicz, L. Stamping out conduct risk in algorithmic trading, KPMG. Available at: Stamping out conduct risk in algorithmic trading – KPMG United Kingdom (home.kpmg) (last accessed 13th January 2022).
2016. About the FCA, Financial Conduct Authority. Available at: About the FCA | FCA (last accessed 13th January 2022).