FINRA algo supervision could be “costly and ineffective”

Tighter rules on the development of execution algorithms in the US could put further cost pressure on the sell-side without improving market stability, experts warn.

Tighter rules on the development of execution algorithms in the US could put further cost pressure on the sell-side without improving market stability, experts warn.

On Friday, the Financial Industry Regulatory Authority’s (FINRA) board approved a series of new rule changes focused on high-frequency trading (HTF) and algorithmic trading. FINRA is responsible for the oversight of securities dealers and exchanges in the US, with the remit to protect investors and ensure markets remain free, open and fair.

The new rules include increased oversight of those responsible for developing algorithms and greater supervision of algorithmic trading itself.

A statement from the FINRA board said it was “seeking comment on a proposal to establish a registration requirement for associated persons who are: (1) primarily responsible for the design, development or for directing the significant modification of an algorithmic strategy; or (2) responsible for supervising such functions.”

The move had been widely expected as US regulators look to gain more control over the way HFT firms interact with the market as a result of growing concerns they are manipulating markets to the disadvantage of investors, however the proposed rule would have a much wider impact, drawing in traditional brokers as well.

Jamie Selway, head of electronic brokerage at agency broker ITG, said the move will create additional costs for the sell-side.

“The number of registered people in our business would grow and there are costs associated with that to pay for them to take tests and additional fees for FINRA,” he explained.

“The costs won’t be prohibitive, but it will be time consuming and create quite an administrative burden for brokers.”

However, the costs are unlikely to be passed on the buy-siders, and will instead just be an additional burden the sell-side needs to bear, according to Matt Samelson, principal at consultancy Woodbine Associates.

“This is unlikely to meaningfully impact costs for the buy-side. Brokers are already competing heavily for their flow and across the whole business the additional cost per trade would be so small that most sell-side firms will simply have to absorb it,” he said.

It is considered unlikely that the US will follow Hong Kong in requiring institutional investors that use customisation tools to adapt with their broker-supplied algorithms to also be registered with regulators, however the sell-side may need to exercise greater oversight of the way their clients are using these tools. 

Costs with no benefit

There are also concerns that these additional costs will provide relatively little benefit. The move is largely seen as a response to the incident that brought down Knight Capital on 1 August 2012, after a rogue market-making algorithm lost the firm US$440 million.

Although the Knight incident highlighted the dangers of deploying algorithms incorrectly, Samelson said the industry has already taken steps to ensure similar accidents do not happen again.

“Brokers do a lot of testing on their algorithms, deploying them in test environments to see how they react to different market conditions. They have tight controls to make sure that any untested code is not accidentally leaked into the trading environment.”

In addition, brokers already do extensive checks on algo development staff to ensure they are highly qualified for the role.

“Our product managers who oversee the algorithms are registered with FINRA, and some of developers as well. This proposal would simply mean more of those development staff have to become registered,” added Selway.

However, the crucial details of how FINRA will implement the rule proposals, which also include steps to increase dark pool and fixed income market transparency and curb some common HFT strategies, will provide greater clarity on how exactly the regulator intends to improve market security through regulation.

Sayena Mostowfi, senior analyst at research consultancy TABB Group, cautioned against making assumptions of how FINRA might apply its registration rules.

“The details of the FINRA proposal will determine their impact on brokers and customers.  For example, we are not yet sure what the registration mechanics would be for someone supervising algorithm development,” she said.

Samelson agreed, “As with all high level rulemaking, the devil is really in the detail and we’ll need to wait to find out exactly how FINRA proposes to deal with increased supervision in this area. But one issue that will need to be addressed is whether regulators have either the resources or expertise to be able to conduct thorough checks on algorithmic trading strategies.”