The regulatory focus on algorithmic trading has never been greater. It is likely to get much more intense with the advent of MiFID II in 2018, which brings with it a raft of obligations for trading firms—in particular for high frequency traders (HFT)—who use algorithms as part of their operations. It is a change not welcomed by the industry.
“We are concerned that the costs of very prescriptive, cumulative, and incremental compliance and governance obligations that will hit small and medium firms hard, leading to a contraction in the number of market participants,” says Mark Spanbroek, vice chairman of the FIA European Principal Traders Association. “This, in turn, could lead to a contraction in the number of market participants and reduce an important source of liquidity.”
MiFID II is not the first rule to focus its attention on algorithmic trading. Germany’s HFT law was introduced in 2013 with the intention of putting algorithmic and HFT within a more robust regulatory framework, ruling over proprietary trading in the same way as for investment firms who hold client money. MiFID II borrows many of the changes introduced by the HFT Act.
First is the explicit definition of HFT and the requirement for HFT firms to be formally registered. The idea is to capture a broad array of firms within the regulatory scope.
“Until recently there was no universally agreed definition of HFT,” says Christian Voigt, senior regulatory advisor at Fidessa. “MiFID II introduces a definition, but in a broad way, and it could impact a few firms who didn’t see themselves as HFT before.”
Registration is not a simple process. As part of MiFID II’s regulatory technical standard (RTS) 6, registered firms will need to adhere to a multitude of rules including showing how software is developed, where it is purchased from, how it is audited, signed off, stress tested, as well as how trading behaviour is monitored and alerts are triggered. The desire to push all HFTs to register under MiFID II has been criticised. Participants say this catch-all standard is likely to force those who might not be engaged in HFT to register as investment firms.
“Any market maker on the exchange is now an investment firm as a consequence of the new law,” says a senior figure at a European exchange who asked not to be identified. “When the German law came in prop firms were not historically regulated. All of a sudden they had to fulfil the same requirements as banks. With the new German law quite a few firms moved to London or have a subsidiary there.”
Diederik Dorst, global head of legal & compliance at Amsterdam-headquartered Flow Traders, says that the overall shift to lit markets under MiFID II is a positive and will help bring transparency for assets where trading is opaque. However, he questions the need to have the specific burden of additional registration under MiFID II for companies that are already subject to similar rules in most developed jurisdictions. Dorst feels that firms such as his own are already adhering to the best practices required by MiFID II as part of their business as usual and feels the changes only add an unnecessary formality to the operational process.
“RTS 6 contains a large number of requirements of organisational and risk controls—but I think that serious organisations should have this already,” says Dorst. “MiFID II codifies what is already in best practice. It borrows from the book of the HFT act—we are already intimately familiar with this because we have been subject to it for years.”
The US has taken a different approach with the affordance of special status to prop firms which do not have client money under management and so are not required to have the same disclosure levels as investment firms—“it would be desirable to have this under MiFID II,” says Randolf Roth, head of market structure and member of the board of directors at Eurex.
Eurex has around 100 investment firms from outside the EU trading on the exchange—however with no decision on equivalence yet there is the fear that these firms will be unable to access the market in future. It is a significant risk.
“Regulators in Europe should address the current uncertainty on authorisation requirements and establish a transitional period,” says Roth. “This would allow both EU and non-EU participants to appropriately adopt the new rules and adhere to the new authorisation requirements in a reasonable timeframe. Especially, firms outside the EU should be able to continue trading in an unimpeded manner, until the EU Commission has decided on equivalence.”
The other major source of alarm around MiFID II is the call for identification of algorithmically generated orders and trading algorithms—so called algo flagging. This has been dubbed as one of the most difficult part of the new rules to implement. The term “algorithm” in the trading sense has never been given a specific regulatory definition before. Identifying exactly what is an algorithm is a huge challenge and will lead to many having to reassess their definitions of their businesses.
“Algo trading is defined very broadly,” says Voigt. “For example, we provide an order management platform which supports, amongst many other order types, simulated stop orders. Because of the broad algo trading definition, we concluded that even a simulated stop order is algo trading. Many people were surprised how the rules classed them as algo traders.”
In the US firms currently only need to say if an order is algorithmically generated or not. The European rules are more detailed with firms having to differentiate between different algorithms. It is a complicated distinction, say participants and one that will not necessarily create any better market structure. From the regulatory perspective, the goal of the algo flagging rule is to increase surveillance and prevent market manipulative trading behaviour, such as “layering” and “spoofing.”
Firms will have to store time sequenced records of their algorithmic trading systems and trading algorithms for at least five years. These records must include information on, for instance, the person in charge of each algorithm, descriptions of the nature of each decision or execution algorithm and the key compliance and risk controls involved in monitoring this.
By collecting vast amounts of data from firms involved in algo trading regulators should, in theory, have more precise historical records of trades making it easier to catch culprits involved in market manipulation and prevent front running or spoofing. But from the trading firm point of view there is a big cost to storing all the data required by the new rules.
“We do have doubts about the vast amount of data that you have to collect and store,” says Dorst. “It is challenging to store the data and requires a lot of technology to do it. If you do so then the onus is on the regulator to do something with the data.”
Worse still, many believe that, being a back-looking exercise, the algo flagging rule is unlikely to stop market manipulation at all.
“To prevent something from happening in the world of fast trading is almost impossible,” says the senior European exchange figure. “Your judgement will be based on something in the past. I don’t expect it will make anything safer because we already do all we can.”
Indeed, Dorst feels that the potential benefits of the algo flagging rule are not a sufficient trade-off for the work involved in keeping and maintaining the data. Nor have regulators devised an efficient plan in how they get the data.
“They could have got the data from trading venues,” says Dorst. “It’s a massive duplication, even tripling the same data in some places. The regulation requires us to store it that way immediately after order submission. This would require highly advanced systems, databases and power for data that just sits there idly.”
The other big area impacting HFT firms under MiFID II is RTS 25—relating to business clock synchronisation. MiFID II will require trading firms’ and venues’ business clocks to be synchronised with Coordinated Universal Time (UTC), allowing the maximum divergence from UTC to be either one millisecond or 100 microseconds. The idea is to deal with the ever increasing speed and volume of financial transactions and produce accurate records like order books. While in theory this is a good idea, the practical implications are likely to prove a further huge complication.
“While RTS 6 seeks a proportionate approach to improve resilience of systems, in comparison RTS 25 lacks such clear benefits compared to what it expects us to do,” says Dorst. “The prescribed requirements are so complex and costly to implement in practice that we honestly think this was overlooked in the rulemaking process, despite input in the consultation rounds for the new rules.”
Dorst says that the requirement for 100% synchronisation at all times under RTS 25 is at this point in time “technically impossible” within reasonable amounts of money and talent being spent on technical infrastructure. It is a conundrum the industry will have to face up to. With only a year left it is perhaps too late to continue arguing and start dealing with the realities of the legislation, harsh as they are.