Under the Water: How Mifid II's Dark Caps Have Changed Trading

If you pour a cup of water into the Atlantic, would the sea level rise as a result? Probably, yes. But because of the size of the ocean it would be almost impossible to measure. 

This is the analogy that Christian Voigt, senior regulatory advisor at Fidessa, uses to describe the difficulties in measuring increases in trading on lit markets since the first data that governs the double volume cap mechanism (DVC) was published by European regulators on March 12. “The problem is that measuring any impact on the lit market is really difficult because the lit market is very large,” he says.

Under the DVC, which was ushered in by the revised Markets in Financial Instruments Directive (Mifid II), if four percent of total trades in a single stock are conducted via a dark pool, that stock can’t trade in that dark pool for six months. If the amount reaches eight percent of volume across all dark pools, then that name is banned from trading on dark venues entirely for six months. 

Trends in equity trading during 2018 have been indelibly changed by the introduction of Mifid II, with a rise in large-in-size, or block trades, the use of periodic auctions, and trading through systematic internalizers (SIs). 

Questions have been asked about how much of this shift can be attributed to the enactment of the DVC. Periodic auctions, for instance, allow traders to move large orders during specified periods while minimizing market impact, and have emerged as a focus of this debate.

The European Commission defines a periodic auction as a “system that matches orders on the basis of a periodic auction and a trading algorithm operated without human intervention.” According to a source at a European regulator who spoke to Waters on the condition of anonymity, what seems to be happening now is some market participants are creating a “new hybrid of auction type” that can be called up on very short notice. The auction period itself can then just last for 100 milliseconds or so.

There has been a significant growth in these types of periodic auctions this year. “There does indeed seem to be a direct sort of association between the double volume cap coming into play, and then certain periodic auction systems being created and gaining some traction in terms of volume,” says the regulatory source. 

According to data from Fidessa, this growth has been from under 1 billion shares traded to around 6 billion by the middle of July. Yet a recent report from the UK Financial Conduct Authority (FCA) found that dark-trading caps were not responsible for this, with growth in the use of periodic auctions observed in both capped and uncapped stocks.    

According to the regulatory source, most European dark pools are based in London. Of course, those venues are also trading shares or other instruments from, for example, France or the Netherlands. “There you can envisage that indeed different regulators may have different views on what to do with periodic auctions,” the expert says.

According to a UK-based market analyst at a brokerage, there are different schools of thought on periodic auctions emerging from regulators. “The FCA is the only one I have seen that has done a data-driven analysis like they presented in that paper. The other regulators haven’t really looked at [it] deeply. They just look at certain features in the periodic auction and think they are copying or replicating what is going on in the dark pools that are capped,” they say.  

Despite the findings of the FCA study, some say certain traders are trying to circumvent the DVC by creating these new types of periodic auction markets.

“The French regulator is kind of concerned about how these periodic auctions work, how the prices are formed within them, and they are kind of doing a further investigation into them,” the analyst says.

Yes or No? 

According to Philippe Guillot, executive director of the markets directorate at French regulator Autorité des Marchés Financiers (AMF), finding a definitive answer to this question would require a vast amount of analysis work. 

“This answer is more subtle than a yes or a no,” he says. In order to properly analyze the impact, Guillot explains, he would need information on all trading flows prior to the original Mifid and then the ability to compare those with trade flows from after Mifid II’s enactment in January 2018.  

The European Securities and Markets Authority (Esma) is conducting a fact finding exercise on periodic auctions, checking how they are being set up and what the features are. The regulator is then expected to discuss its findings with member states and decide whether there are any additional regulatory or supervisory steps to be taken, and the outcome of that process at this point is very much open.

Another significant contributor to changing trading patterns has been the prohibition of broker crossing networks (BCNs), which have instead been replaced largely by SIs, entities used to transact client orders against a broker’s own capital. According to Guillot, the reason for banning BCNs was to bring back on more lit venues those flows that have been increasingly confined to smaller and smaller crosses in those BCNs. 

“It seems that some of those flows, moved from BCNs to SIs,” he says. “But I do not have the order books on SIs and so on, and I cannot tell you today with absolute certainty that the flows that were going to the BCNs yesterday, are now going to the SIs. If you wanted me to make that assertion, I would need proper pre-Mifid I data to compare to pre-Mifid II data. Mifid II gives me more data than Mifid I, so I will never have comparable data, but if I wanted to be able to do that, I would need comparable data from the order books, between Mifid I and Mifid II.”   

It is also difficult to judge how much of the SI volume is actually firm, addressable liquidity, traders say, meaning that even the data gathered under the enhanced requirements in Mifid II does not always paint a truthful picture.

“Under the SI regime, trades are often tagged as SI whether they are addressable or not. So while SI volumes have ostensibly increased, much of it may not be real, addressable liquidity,” says Joelle Tarrant, global head of market structure at HSBC. 

On the lit markets it would be even harder to tell what impact there has been, even if the complete Mifid I order books Guillot referred to were available, along with a computational platform powerful enough to accurately parse, compare and analyze the results. That’s not to say data was missing under Mifid I, but rather, it was affected in a major way by one particular event—the global financial crisis. 

“The same problem happened with Mifid I if you will remember,” says Fidessa’s Voigt. “So when Mifid I was done, after three years it was supposed to be reviewed, right? But the problem was that when the Mifid I review was supposed to take place, we had the global financial crisis. So all of a sudden nobody was able to do any statistical analysis anymore, because you had this massive financial crisis, which nobody blamed on Mifid I, but obviously anything done under Mifid I was heavily impacted by it.” 

Yet the answer to this puzzle—and indeed, the puzzle itself—is far more complicated than it seems. The very construction of the DVC is shrouded in a sense of mystery that makes judging its utility difficult.   

Dark Forces

A key objective of the DVC is to move more trading from the dark to the lit markets, ostensibly for the sake of market transparency. But there are valid reasons for trading in the dark, with a long history attached. 

“In the old days where you only had humans, you know what a dark pool was? Effectively, it is a pocket of the traders. It was never written down anywhere. But in effect, it was very similar trading practices, because they knew if they had a very large order they can’t immediately disclose it to the market because it would harm the client,” says Voigt. 

But dark pools really began to take off in Europe when Mifid I was introduced in 2007. At that point, trading was largely confined to national exchanges and speed was less of an issue. Mifid I allowed for an explosion of trading venues, such as multilateral trading facilities, BCNs—and of course, dark pools. 

“The fact that you see smaller trades in the dark pool is just a function of how electronic trading has developed,” says the market analyst. “And you can use algorithms to look across a lot of dark pools to maximize access to liquidity and control market impact. So I think it is just the way that algorithmic trading has developed, which has adjusted how people trade within these dark pools. So rather than posting a large order there, you post lots of small orders across different dark pools, to make sure you can capture all the available liquidity. “ 

But even though the point of the DVC is to increase transparency, concerns remain about its construction. The DVC is not a very elegant way of trying to control dark pool trading, the analyst says. 

“There was no real analysis done beforehand on why four percent and eight percent were the best numbers to use,” they say. “I have never seen a European regulator stand up and say ‘well we chose eight percent because we looked at how much dark trading would be detrimental to the market. They just seem to have picked those numbers out of thin air. I also looked at other regions—Australia and Canada—where they have taken a slightly different approach to controlling dark trading and it is all based around whether dark pools can provide you with a better price than you would with trading in the lit.”  

While that is not a perfect means of analyzing the impact of dark trading, the analyst suggests it is perhaps “a more thoughtful way of doing it,” rather than simply introducing hard caps. The emphasis, he says, should be on execution quality rather than arbitrary numbers. Even the regulators agree. 

“When the parliament voted on Mifid II, those caps of four percent and eight percent were written in the text of Mifid II. So it is not something that comes from Brussels nor from the regulators,” says the AMF’s Guillot. The regulatory source concurs, describing it as a political issue.

Indeed, few people seem to have an idea of exactly where the numbers came from, or why they have been widely accepted, outside of political diktat.

“I have asked this question a million times, and so far I have not gotten a satisfying answer,” says Fidessa’s Voigt. “It is an absolutely valid question. And for the life of me I don’t know. All I can tell you is there was a draft document published that initially said, I think, in the beginning, 5 and 10 percent. And then it was miraculously changed to four and eight some time later.” 

Some market participants are unhappy with the DVC because of the price they have to pay based on these seemingly arbitrary numbers. “You can imagine that certain trading venues that operate dark pools are very unhappy with this because it means the amount of volume they can do is effectively curbed,” says the regulatory source. “And it is also curbed in a kind of unpredictable way because a suspension can hit at any given time. And then you have to adapt and then the volume that you do in the next month can be restricted.”

Will the DVC Be a Success? 

When it comes to the long-term impact of the DVC, Guillot says regulators have always been very bad at predicting the future. He says if he had been good at it, he would have worked in the stock market instead of being a regulator. 

A theme that remains consistent is that most see the DVC as a political tool, rather than a means of ensuring best execution or enhancing transparency.

“I would say there was a policy objective for Mifid II and the dark caps to make more trading happen in the lit markets. And our data shows that has been achieved,” says Duncan Higgins, head of electronic sales at ITG. “I don’t agree with the policy objective, but that is another matter.” 

Fidessa’s Voigt says the legislation has been a compromise. “I hope it is just going to be an annoying footnote. Because the way I see the DVC is that it was a political compromise done at a time when politicians were desperate to get a deal and an agreement across very diverse parties,” he says. “If you compare the US with Europe, what you need to consider is in Europe effectively you have 28 countries at the moment that all need to agree on a financial market regulation. So you have 28 diverse views that you need to bring under one hat. And I don’t think anyone really liked the DVC.”  

Despite grousing over the nature of how the DVC was put together, however, the regulatory perspective—leaving aside, perhaps, the hedging from the AMF—seems to be that the system as constructed is working well so far, inasmuch as the calculations are being produced consistently and the questions are being asked about the wider impact.

“From our discussions with regulators and from public dialogue at conferences, regulators seem relatively confident that DVCs have been successful in their implementation,” says HSBC’s Tarrant. “They are now examining secondary ramifications, such as rises in periodic [auctions], and rises in SI volumes.”

But from the markets perspective, there is still probably bit of a hesitation around the benefits that the DVC can bring.