Algorithmic trading: trends and existing regulation
Algorithmic (ALGO) trading refers to trading in financial instruments where a computer algorithm (the “execution algorithm”) automatically determines order specifications, such as when to initiate the order, the price or quantity of the order and how to manage the order after it is submitted. This takes place with limited or no human intervention. High frequency trading (HFT) is a type of ALGO trading where more complexity and speed are usually involved.
ALGO trading has certain advantages compared with manual trading. It makes trading processes more efficient by reducing labour and other related costs. It also enables large volumes of data to be analysed in very short time frames. For some forms of trading this is especially appealing. For example, arbitrage trading requires orders to be identified and executed quickly, as arbitrage opportunities are short-lived in the market. Moreover, trading based on the identification of trends and statistical metrics can be performed faster and identified more precisely by a computer.
ALGO trading has been growing steadily since the early 2000s and, in some markets, is already used for around 70% of total orders. This growth has been facilitated by technological developments, such as increased computing power, reduced storage costs and the implementation of artificial intelligence and machine learning techniques. In the banking sector, cost considerations, competitiveness and profitability are key incentives to trade using algorithms.
But ALGO trading also entails risks stemming from potential failures of algorithms, IT systems and processes. In recent years, a number of major ALGO trading failures have resulted in substantial losses, fines and reputational damage for credit institutions and investment firms.
These risks, and the rapid expansion of ALGO trading, have led to increased attention to the applicable supervisory and regulatory framework.
The Markets in Financial Instruments Directive (MiFID) and its delegated acts introduced comprehensive requirements to mitigate the risks for credit institutions and investment firms engaging in ALGO trading. Similar requirements were also introduced for trading venues. The requirements cover, among other things:
- governance (including an algorithm approval process commensurate with the board’s risk appetite)
- staffing (qualifications, skills and size of the workforce)
- outsourcing of IT
- testing of the algorithm’s compatibility with the systems of the trading venue and market access provider, and stress-testing of the algorithm itself
- business continuity arrangements to deal with disruptive incidents
- resilience (“kill” functionality – shutting down the trading system in the event that the trading algorithm is running erroneously)
- surveillance (automated surveillance to detect market manipulation)
- pre-trade controls (order entry), real trade controls (market monitoring) and post‑trade controls
- security (physical and cyber security, limitations on staff with critical access rights to the systems)
Oher MiFID provisions may also help prevent or mitigate the potential negative effects of ALGO trading on financial markets. These include minimum obligations for institutions pursuing market-making strategies based on algorithms to provide liquidity to the trading venue where they operate and limitations on the ratio of unexecuted orders to transactions.
National market supervisors have been given the mandate to enforce the MiFID requirements vis-à-vis investment firms and credit institutions engaging in ALGO trading. In parallel, the mandate of the ECB, in its role as single prudential supervisor for significant credit institutions in the euro area, consists of preserving the safety and soundness of these institutions. For that purpose, the ECB is required to enforce the Capital Requirements Regulation (CRR), which imposes capital requirements on all institutions performing trading activities. These requirements apply to all trading forms, including ALGO trading, and are mainly based on the market risk or volatility of the trading portfolios. Moreover, ALGO trading incidents could also be linked to the prudential definition of operational risk in the CRR because they are often linked to failures of systems and processes. Therefore, ALGO trading risks would also be considered indirectly by the ECB, within the broader supervision of operational risk. In practice, MiFID requirements also play an important role in reducing operational risks.
Despite the many advantages of ALGO trading, individual system failures resulting in significant losses have highlighted the need for institutions to develop specific risk management capabilities and for regulators and supervisors to monitor the risks that it poses. The new MiFID requirements enforced by national market supervisors will reduce the potential operational risk embedded in this type of trading. Prudential supervisors, for their part, will remain vigilant as to the risk management implications and governance of ALGO trading.