Provisions that would have required a minimum 500 millisecond pause on incoming trades have been dropped, alleviating some of the concerns about the potential impact on liquidity, and trading shifting to other jurisdictions.
However, participants engaged in high-frequency algorithmic trading will need to be licensed, and trading facilities must limit the ratio of unexecuted orders to transactions in their systems.
Trades generated by algorithms will have to be flagged. The tag will contain an identifier code linking the trade to a specific algorithm to allow regulators to trace and crack down on trading strategies that are “abusive” or pose risks to the market.
Providers of direct electronic access will be responsible for ensuring clients’ trades executed through their systems comply with the new rules.
The changes – which the EU estimates could cost as much as €732 million to implement – come despite widespread agreement that trading technology has enabled wider participation in markets, increased liquidity, narrowed spreads, reduced volatility and enhanced execution of orders.
“The arguments that high-speed trading is in some way damaging to markets have all now collapsed,” says Simon Gleeson, partner at Clifford Chance. “Nobody has ever managed to demonstrate that high-frequency trading poses any particular risk to participants. It is not necessarily high risk, yet it is capable of being extremely risky. But then the same is true of real estate development lending.
“Frankly, something that goes wrong once every five years is not the most inherently risky activity imaginable.”
Having escaped mandatory speed limits and resting periods imposed by law, elements within the sector are proposing to do just that of their own accord.
In April, EBS, one of the two big foreign exchange trading platforms, called for “first in, first out” execution of orders to be jettisoned, arguing that it gives an unfair advantage to participants with the highest computing-processing power and fastest servers.
ParFX, a new platform backed by large European universal banks which launched earlier this year, already operates a system of random pauses on incoming orders, saying it ensures fairness.
“This is a very important development that demonstrates that FX market practitioners are taking active steps to find a solution around latency,” says Stephane Malrait, chair of ACI FX Committee. “The main difference with the equity market is that the FX market needs to serve many different client types across the globe.
“To be able to deliver a global service, banks need to access liquidity without being impacted by latency or location of their technology. The idea to implement randomized pauses for incoming orders is a good way to solve that issue.”
Attention has focused on incidents such as the one involving Knight Capital last summer when a rogue algorithm landed the firm with billions of dollars of unwanted shares, but algorithmic high-frequency trading is increasingly employed in FX.
According to FX settlement system CLS, average daily global foreign exchange turnover stood at $5 trillion in April. That is an increase of $1.7 trillion in the past five years. The Aite Group estimates that by the end of next year more than 25% of FX trade will be algorithm-driven.
And while the new regulations do not apply to spot FX, trade in other FX instruments, which is worth at least $2.5 trillion a day, will be impacted.
As market-makers, banks make up the largest single category of participants in the FX market, but other financial institutions are catching up.
Non-reporting banks, hedge funds, pension funds, mutual funds, insurance companies and central banks account for three-quarters of the non-spot market, according to the Bank for International Settlements.
The compromise rules appear to provide a strong regulatory framework that reduces risk while allowing the market to benefit from the latest advances in trading technology, but critics remain unconvinced.
“The lesson we have all been taught painfully, and to our cost, over the years is that with all regulation the question is: will the impact that it has have sufficient economic consequence to justify a change in behaviour,” says Gleeson.