Automated Trading & Distribution
Summary
Electronic and Algorithmic (Automated) trading has expanded rapidly from its early roots in equities, extending to FX and, increasingly into Fixed Income.
•Automated trading brings significant benefits, including cost reductions and operational efficiencies. However the nature of automated trading activities has radically altered the transactional processes and requires very different skills and expertise across the trading floor.
•The advancing use of AI on trading floors is increasing the remoteness of human involvement in the trading and distribution process.
•Additionally, the possibility of alternative types and methods of conducting market abuse (deliberately or inadvertently) add further complexity and requires close monitoring.
•Consequently global regulators expect financial organizations to install appropriately robust infrastructures to suitably manage and oversee these new risks. There is a strong emphasis upon the individual responsibility of (named) senior management to establish and then maintain an appropriately robust control framework – consistent with requirements of the Senior Management and Certification Regime (SMCR) and Global FX Code of Conduct.
•FMCR is a specialist front-office consultancy. Our consultants have senior practitioner expertise and experience in electronic trading across Equities, Commodities, FX and Fixed Income. This knowledge enables us to undertake a detailed review of current and intended practices and recommend any enhancements to the control structure for automated trading and distribution.
Background
•Automated trading is defined (by the UK’s FCA) as:
•Where a computer algorithm automatically determines individual parameters of orders (e.g. timing, price or quantity).
•Where there is little/no human intervention in the trading process.
•However there is no uniform definition of “algo” trading across institutions.
•The sharp growth in automated trading over the past few years has unquestionably brought some notable benefits:
•Reducing conduct risk, due to reduced levels of human involvement.
•Reducing human error (e.g. fat finger) risk.
•Reducing operational, personnel, and distribution costs.
•Creates an infrastructure facilitating higher trading volumes.
•Simultaneously, the nature and complexity of automated trading presents substantially different challenges from the traditional voice-led trading. This has created concerns amongst global regulators, including the UK’s FCA and PRA.
•The UK regulators require “a named person” to be created at every institution, responsible for how and where the algorithms/AI are employed. This ties in with the requirements of the SMCR.
•In reality there exists a spectrum of Automated Trading, spanning minor automation and (essentially human-led), through to AI, whereby the system decides when and how to place/cancel orders.
There are several types of risk that have features specific to Automated Trading;
i.Governance
ii.Technology
iii.Model
iv.Systemic
v.Conduct and culture
i. Governance risk:
•The technology-led nature Automated trading exposes organizations to a broad range of additional risks;
•Inadequately documented policies and procedures, creating the danger of “knowledge-gaps”.
•A lack of clear ownership and management oversight resulting in potential to “pass the buck”.
•A lack of understanding of the challenges created by automated trading at Senior and Board level.
•Insufficient depth of technical expertise and practitioner knowledge in Risk Management, Compliance and Internal Audit functions.
•AI brings notable additional risks, where the system is the sole decision-maker in some circumstances.
ii. Technology risk:
•It is expected by the regulators that firms using Algo trading will have “knowledgeable and qualified” staff, conducting real-time monitoring and with the authority to take radical actions, including operating a “kill switch”.
•Algo trading holds multiple potential technology risks, including:
•Inadequate system planning/testing and IT systems incompatible with each other.
•Lack of planning and testing for operational resilience in the event of unexpected “disaster” events
•Constant dangers of Cyber attacks.
iii. Model Risk
Model risk has been prevalent within traditional trading business for many years, including the poor documentation of underlying assumptions, lack of comprehensiveness of the data quality and incompleteness of data sets etc.
However there are model risks around specification, implementation and usage which are more specific to Automated trading.
These include:
•Lack of intuitive understanding/expectation of the model outputs in varying market conditions, potentially due to a lack of understanding/inadequate checking of inputs and coding.
•Embedded logic which could lead (inadvertently or otherwise) to market-abuse behaviours.
•Incorrect interpretation of outputs, by traders & risk managers with inappropriate backgrounds/lack of understanding.
iv. Systemic risk
The nature of trading creates the capacity to amplify systemic risks, as recognized by global regulators.
The systemic risks include:
•Violent swings in prices, and more unexplained volatility, e.g. due to “herding” and “panic” characteristics.
•Lower liquidity in volatile markets, as programmed systems temporarily withdraw from the marketplace.
•Liquidity and concentration risks arising at particular moments, from Algo – related trading, can lead to “bunching” into similar trades, thus causing liquidity and concentration dangers.
v. Conduct and culture risk
•The FCA expects firms to meet behavioral conduct objectives:
•supporting fair and orderly markets
•promoting effective competition
•achieving fair client outcomes
•Automated trading broadly reduces conduct risks. However, it does potentially open up other, less recognized or understood, conduct risks; optimal outcomes may not be consistent with conduct rules.
•There is a need to ensure against programming in (deliberately or otherwise) behaviours which could potentially cause market abuse (manipulation) outcomes
•Controls are required around programming of the sales/distribution activities (often occurring via a number of outlets), to ensure that the pricing is demonstrably “fair” and consistent for (eventual) clients.
•Methods can be used to reduce conduct risk (e.g. surveillance systems, training) but there is also need to guard against creating a “terror” atmosphere on the trading floor e.g. surveillance systems throwing out “false positives”
Regulators consistently emphasize the personal responsibility of senior management in exerting control over the Automated activities.
Identifying and addressing poor practices
The FCA has previously identified a number of areas of focus and provided examples of poor practice;
•Defining algorithmic and AI trading:
- firms failing to adequately demonstrate how and where Algos/AI are used within the firm.
•Development & testing:
- firms unable to provide effective MI to Senior Management, including on risk parameters, surveillance & training.
•Risk Controls:
- firms have not been able to demonstrate a structure and tailored approach to risk controls.
•Governance & oversight:
- senior managers are expected to take personal responsibility for the Electronic and Algo activities.
•Market Conduct:
- firms are expected to constantly consider all potential types of market manipulation which can result from the use of Algos.
Establishing a robust control and governance framework;
The nature of automated trading requires a control and governance framework which recognizes the different characteristics which this style of trading brings including ensuring;
•The various risks are being appropriately captured, evaluated and documented, on a continuous basis.
•Processes and procedures are established, maintained and robust; vital given that the pace of change is extremely fast – this incorporates the full mapping of end-to-end flows.
•There is sufficient depth of understanding and expertise within the functions and at each level of management seniority within the organization.
•The control of automated activities is underpinned by the 3 Lines-of-Defence (3LoD) model, consistent with the oversight of “traditional” trading operations.
•Automated trading is overseen by a global Risk Committee, in recognition that there are unique aspects which require a differing approach to the firms’ control framework and supported by appropriate specialist sub-committees.
•The risk governance framework is thereby adequate for approval by the Board.
Role of FMCR - Automated Trading Review;
FMCR is able to provide practitioner reassurance to Senior Managers regarding the robustness of Trading floors, framed around the FMCR Trading Floor Health Check.
•Undertake a forensic independent review , carried out by former ex-practitioners with expert insight.
•Conduct analysis and back-testing review of automated trading models
•Consider potential for Market Abuse risks, including reviewing the automated pricing mechanisms to end-users.
•Perform a “technical audit” assessing;
•Methodologies
•Processes and procedures
•Documentation
•Management oversight.
•Assess the completeness of mapping of end-to end flows
•Ensure the automated trading offering is compliant and current in accordance with regulators expectations.
Conclusion
•The continued rapid growth of Automated trading and distribution is undoubtably bringing substantial benefits, not least from cost and volume perspectives.
•However, in many aspects it is a very different type of operation to that traditionally carried out on trading floors. These differences need to be recognised, addressed, fully documented and constantly reviewed.
•This has been stressed by global regulators, who constantly emphasize the “personal responsibility” of management.
• In order to establish and build a robust control environment, it is important that there is deep understanding through the organization of the components of the Automated trading and distribution environment
FMCR has the specialist skills, expertise and experience to undertake reviews, make recommendations and assist in the implementation and on-going maintenance, including training and Board-level education.