European Dark Pools: Quality, Control and Change
While the demand for non-displayed liquidity in Europe remains strong and the growth of European dark pools continues, there are pools and then there are pools. The quality of liquidity in a dark pool depends not only on the participants, but also on the quality control the dark pool operator is able to exert. Moreover, as Christian Hesse, Senior Quantitative Strategist for Autobahn Equity Algorithmic Trading at Deutsche Bank explains, there is the issue of pending regulatory change to consider.
Senior Quantitative Strategist,
Autobahn Equity Algorithmic Trading,
Since the adoption of the Markets in Financial Instruments Directive (MiFID), European dark pools have seen steady growth. Recent TABB Group estimates (see Figure 1) indicate that at the end of February 2012 there were 38 different dark trading venues in Europe accounting for about 8% of all equity trading. By contrast, the growth of dark pool trading in the US has been levelling off over the past few years with the number of venues and market share remaining fairly stable at 54 and around 12%, respectively.
Driven by the Buy-Side
In Europe, the growth and popularity of dark pool trading has been driven by solid demand for access to non-displayed liquidity from the buy-side. In some cases, institutional investors have made their willingness to do business with a financial institution conditional upon the provision of access to its internal liquidity through a broker operated dark pool. This buy-side insistence on pool provision has been followed up with actual trading activity. For example, Deutsche Bank's dark pool, SuperX, which was launched in early 2010, since August 2011 is consistently in the top three of the 18 European dark trading venues appearing in a monthly survey by Rosenblatt Securities. For December 2011 SuperX was ranked the largest venue by turnover for non-displayed liquidity in Europe; in March 2012 SuperX was the second largest European dark pool.
Dark Pool Quality
Notwithstanding the relative success of some dark venues, there is concern that the apparent decline of average trade sizes across dark pools may be undermining the value of dark trading for buy-side participants. In Europe, dark trade sizes are usually expressed in terms of the notional value of the executions converted into Euros, rather than the number of shares. This simplifies averaging and reporting across stocks and markets, but exposes trade size measurements to fluctuations in foreign exchange rates and share price, as well as large differences in company valuations.
Looking at all European stocks over the past year-and-a-half, we compared the monthly average notional values of trades in 15 of the European agency, broker or exchange operated dark pools registered as a Multilateral Trading Facility (MTF) - for which data is publicly available - with those of trades on venues with displayed order books. Figure 2 shows an overall decline in both lit and dark trade sizes, which may be partly attributed to challenging conditions in European equity markets. Moreover, we see that the ratio of monthly average dark and lit trade sizes has been fairly constant. So it seems that dark pool trading has generally been "larger-than-lit" trading, and that buy-side concerns about trade sizes in European dark pools are perhaps somewhat exaggerated.
Nevertheless, as the number of dark pools has grown, so has the number and variety of market participants trading in them. Increasing market fragmentation and access to dark pools by liquidity seeking and conventional execution and routing algorithms could indeed be putting downward pressure on dark trade sizes in addition to trends seen in the lit markets. If this is the case, then it is all the more important for buy-side participants expecting to trade dark in size to manage their dark pool trading to favour interaction with dark liquidity meeting their objectives.
Figure 1: The growth of dark trading in the US and in Europe
Source: The TABB Group
Selective Sourcing of Dark Liquidity
Irrespective of whether participants access dark pools directly, or via an agency execution algorithm, they have full control over the timing and size of their order submissions and moreover can set limit prices and specify minimum acceptable quantity (MAQ) criteria that any counter party has to meet in order to be eligible for executing against their order. Limit prices and MAQ settings effectively provide high level (and fairly static) control over the dark trade size in terms of notional value.
A dark liquidity seeking execution algorithm such as Deutsche Bank's SuperX Plus strategy provides additional dynamic controls over MAQ and limit price settings which are data driven and adapt to prevailing market conditions. Stock and venue specific MAQ settings help to optimize the algorithm's dark pool routing and allocation decisions, while the Dynamic Return Model (DRM), essentially a form of fair price model, continually adapts order limit prices in order to guard against adverse selection and prevent executions when the price can be expected to revert to more favourable levels in the short term.
Controlled Interaction of Dark Flows
The regulatory status of a broker operated dark pool can have a major effect on the quality of the liquidity experience that it is able to offers to participants. For example, Deutsche Bank's SuperX is classified as a Broker Crossing System (BCS) rather than an MTF dark pool or Systematic Internaliser (SI). At present, the operator of a BCS dark pool is free to exercise a certain amount of discretion in managing the interaction of different types of flow within the pool. In this context the ability to analyse and monitor the characteristics and composition of dark pool liquidity, reflecting the interactions of flow from different types of participants, is essential.
For example, a large order left resting in the pool will see a particular execution profile, which in terms of timing in relation to price changes on lit markets will be fairly random. On the other hand, a participant trading using a systematic intraday strategy will exhibit a very different profile. Equipped with this information, it becomes possible to structure the interactions among participants in a BCS dark pool so as to best meet the expectations and trading objectives of different client groups. For instance, particular clients might exercise their option not to interact with certain other types of participant, e.g., in SuperX what we term a "long-term alpha" client can elect not to interact with flow from a "short-term alpha" client.
Our dark liquidity analyses then allow us to illustrate to individual clients how these managed flow interactions do in fact help to achieve what they intended at a fundamental level. Furthermore, it enables us to show clients the properties of flow in other parts of the pool they are not interacting with, but which might add value from their perspective given their execution objectives. We are looking into modifications to our suite of execution algorithms that would allow clients to dynamically specify the selection of types of SuperX liquidity they interact with to best suit their trading objectives in different scenarios.
Figure 2: Trade Sizes for European Dark Pools and Lit Venues
Source: Thomson Reuters, Deutche Bank AG London
Interaction of Agency and Principal Flows
Another benefit of a BCS dark pool is that algorithmic trading flow can be allowed to interact with the broker's principal flow (not to be confused with proprietary trading activity). This adds considerable value in that principal flow also ultimately originates from a client investment decision - the only thing that differs is the mechanism the client has used to action that decision. On the one hand, the client has entered it into an algorithmic order ticket. On the other, the client has called a bank program/cash trader to request a price and an immediate fill (in which case the position is then on the bank's risk book and traded out of at the bank's discretion).
In the case of Deutsche Bank, both these client-initiated flows can currently cross freely and safely within SuperX, which is beneficial because it increases the depth of liquidity in the pool. Moreover, principal and agency flows interact on a level playing field because the bank's equities traders use the same suite of execution algorithms as our Autobahn Equity clients.
Adapting to Regulatory Change
Proposed changes to the regulation of electronic and algorithmic equity trading in Europe currently being debated under the MiFID review are likely to have significant impact on broker operated dark pools. While the details have yet to be finalized it is probably reasonable to assume that BCS dark pools will eventually be forced to classify as an SI, an MTF or an Organized Trading Facility (OTF). An SI allows independent price formation but incurs a pre-trade transparency obligation to stream quotes and prohibits agency-agency flow interactions. An MTF dark pool, whose prices are linked to a reference exchange, would reduce operator discretion over managing participant interaction. An OTF affords more flexibility in managing the participant interaction, but (the pool operator's) principal flow is prohibited from interacting with any agency flow.
It is certain that the business and operating models for broker dark pools will change under the new regulatory regime, as will the dark liquidity landscape. As a broker and the operator of one of the largest European dark pools, Deutsche Bank is well positioned to adapt to pending changes in regulation, and is committed to using its experience and expertise to continue to provide and source the particular kinds of dark liquidity that its buy-side and other clients are demanding.