Skip to main Content
Site Search

Advanced Search

  • Mondo Visione
  • Mondo Visione - Worldwide Exchange Intelligence
Member Login

Member Login

Forgotten your password?

MiFID update 2008 and liquidity pools, dark and light

Date 29/10/2008

Robert Barnes, FSI

Managing Director, Equities, UBS Investment Bank and Chairman, Securities Trading Committee, London Investment Banking Association

Effective 1 November 2007, MiFID, the Markets in Financial Instruments Directive, is European regulation designed to enhance investor protection, promote competition and increase transparency across the financial services industry.

Expect more. Expect more competition, more complexity and more change, with opportunities for those delivering the highest standards. Why is that?

Because MiFID impacts three relationships:

  1. How buy-side firms interact with their clients
  2. How sell-side firms interact with their buy-side clients
  3. How exchanges and brokers interact

Changes to these relationships will lead to the following consequences:

  • Competition will reduce exchange and central counterparty (CCP) fees, starting with cash equities. Why does this matter? Because lower exchange fees increase liquidity, help reduce market impact, and ultimately improve investment performance; and these changes are good for the market.
  • The market landscape will become more complex. With competitive new entry, there will be more choice of execution venues and related entities which will require more sophisticated tools and methods to manage.
  • Buy-side firms embracing change and building on unbundling trends can become more efficient. As buy-side firms have new best execution obligations to their clients, the potential for added importance derives from how effectively buy-side firms choose between executing brokers using MiFID criteria that excludes research, while separately rewarding value-added research, for example, through Commission Sharing Agreements (CSAs).

Conclusions about the impact of MiFID revolve around how the landscape will change. Volumes will grow due to lower exchange fees and upgrades to faster matching engines. Smart order routing will arrive in Europe to pool, in a virtual manner, the multiple physical liquidity puddles resulting from the new venues. Order flow and new business will concentrate to those most capable in an increasingly fierce technological and commercial arms race.

MiFID sets the regulatory framework encouraging competitive new entry and entrepreneurial opportunity. Realising this opportunity arises from understanding and embracing impending change to market structures.

What do we mean by market structures?

Market structures comprise the rules and institutions that determine competition among trading platforms. This definition encompasses the framework for interaction, including exchange fees, which ultimately shape order execution strategies. The focus includes external factors that impact business and operating models driving opportunities to grow revenues and to reduce costs.

Competition enabled by MiFID and encouraged by the regulators

There is a symbiotic relationship between exchanges and their members, including brokers. One can see the brokers as a free sales and distribution arm for exchanges, for example piping order flow onto exchange orderbooks as brokers sign direct execution client or when brokers hedge blocks in smaller pieces on-orderbook.

As trading flows increase (for example, through competitive broker automation) and exchange internal unit costs fall (as should be the case with fixed cost platform providers), exchange members have a reasonable case for exchange tariffs to improve. Competition can lead to lower frictional costs.

According to the World Federation of Exchanges, the value of cash equities traded around the world in the most recent year 2007 was equivalent to USD100tr with approximately half (USD47.7tr) represented by Americas. The lower level of value traded in Europe (USD31.4tr) presents an interesting opportunity, as economists indicate that the American and European economies are similar in size. Reducing frictional costs of trading, i.e. reducing exchange and post-trade fees, may stimulate more liquidity in Europe.

Figure 1: Cash equities value traded by region

Source: WFE, single counted

Figure 1 shows 18 years of data from the World Federation of Exchanges with the top right point highlighting the USD100tr total value of cash equities traded in 2007. While there was a peak in 2000 just before the bursting of the ‘internet bubble’, the overwhelming trend of the data is that values trading on exchanges are increasing.

Brokers directing this flow to exchanges operate in an increasingly competitive environment. To compete at the top, brokers must innovate.

Fast technology is the enabler, algorithms the logic, and automation by brokers speeds the process. Clients benefit from brokers offering faster execution, potential price improvement and better service. The positive result is increased market activity.

A focus area for order execution strategies is minimising market impact, for example by slicing orders into ever-smaller sizes before deploying them, intelligently, onto exchange orderbooks. The consequence for exchange orderbooks is a trend of increasing number of bargains and smaller average size per trade as shown clearly in the next table.

Table 1: London Stock Exchange orderbook 1998-2007

  Order book average size (GBP) Order book number of bargains
2007 14,908 134,150,345
2006 19,362 78,246,367
2005 20,463 51,415,546
2004 21,472 40,771,163
2003 21,739 32,897,427
2002 28,126 23,839,550
2001 41,283 15,750,253
2000 61,749 8,594,471
1999 63,020 5,374,520
1998 58,508 3,583.128

Source: Factsheets and News

Table 1 summarises data from the London Stock Exchange (LSE). The trend is one of increasing number of bargains and lower average bargain size. This trend is broadly similar for all exchanges, with electronic orderbooks processing activity from brokers competing with other brokers in a fiercely challenging environment.

The timetable displayed coincides with brokers deploying increasingly sophisticated order execution engines. In 1999, the ‘internet bubble’ neared its peak, and approximately 5m bargains by number matched on LSE’s orderbook across all members. By 2007, orderbook trades had grown to more than 134m. Looking only at the growth in number of bargains, one never would have guessed there had been a ‘bear’ market occurring from 2000 to 2003. The table also shows the average bargain size falling from GBP63K to less than GBP15K in 2007.

By multiplying the average size by number of bargains, one can see that the overall value traded grew from 1999 to 2007 by approximately six times the number of trades grew by approximately 25 times. Exchanges benefit from these trends as they charge tariffs on a combination of number and value variables for processing trades through what are broadly fixed cost platforms.

As a fixed cost platform processes more activity, the internal unit cost of production falls which is good for the exchange’s profitability. The benefits of economies of scale lead to more internal efficiencies at the exchange, and this is shown clearly in the example provided by Deutsche Börse summarised in Figure 2.

Figure 2: Economies of scale: Lower unit costs in the cash and derivatives markets

Source: Deutsche Börse 2002 annual report, page 24

In Figure 2, Deutsche Börse shows how its internal unit costs of processing a trade on Xetra, its exchange orderbook, fell from a normalised 100% in 1998 to 18% by 2002. This trend of reducing internal unit costs is broadly true for all exchanges with electronic orderbooks over the last few years. Although the chart stops at 2002, one would expect the reduction in unit costs to continue as orderbook activity has grown dramatically since then.

Meanwhile, exchange fees to members have not reflected these economies of scales. In fact, the public statements from some of the exchanges that had attempted a transformational merger with the London Stock Exchange, for example, simply suggested intentions not to raise exchange fees (Macquarie, Nasdaq) or in some cases promised the potential sharing of synergy benefits with users to the tune of approximately 10% cut in exchange fees (Deutsche Börse, Euronext). LSE, still independent, itself announced in January 2007 a selection of tariff cuts that will eventually be worth approximately 10%. Such offers, while welcome in principle, are underwhelming in context.

With the introduction of cash equities central counterparties (CCPs) to European markets in recent years, a new layer of frictional costs has appeared on top of exchange fees for matching trades. So even where functionalities, such as netting, physically reduce the number of settlements, the current European exchange and CCP tariff structures mean that exchange and CCP fees for processing business have increased, contrary to expectations. This suggests a compelling need to review tariff structures.

Some European exchanges and CCPs have engaged positively with the user community, for example through their own established user advisory groups or via the industry’s representatives at the London Investment Banking Association (LIBA). This has led to new tariff structures which do yield lower marginal and average unit costs in some cases. The magnitude of the overall trend, however, of increasing number of bargains and lower average bargain size mean that the frictional costs of trading remain significant, and exchange and CCP tariffs are material variable costs to brokers which becomes increasingly important to the market as business scales.

There is widespread belief in the market that additional value for money has not increased commensurately with exchange volumes and revenues. The industry view, originally expressed through LIBA and increasingly widely supported by other trade associations, is that exchanges and CCPs should address three issues regarding tariffs: headline cuts (to reflect the significant contributions already made by members to the platforms), incentives for incremental flow (such as volume discounts, caps on aggregate fees, plus other creative mathematical ideas), and simplification of invoices.

What is the context for these consistent requests? Exchange members, including brokers, start with goodwill towards exchanges and recognise the positive symbiotic relationship between exchanges and their members. Ideally, the way to grow business further is to work together with the incumbent exchanges and CCPs in a spirit of entrepreneurism and partnership. The motivation for reduction in frictional costs, particularly European exchange and CCP fees, is to increase liquidity, lower the market impact of pro-competitive broker order execution strategies, and ultimately increase investment performance for buy-side clients.

The reality today is that brokers have little influence with most exchanges and post-trade providers such as CCPs, beyond the goodwill the exchanges and post-trade providers offer, and this is particularly true in Europe. Given the current landscape and activity trends, one can understand why the incumbent exchanges and post-trade providers prefer to preserve their privileged positions and yields from current tariff structures.

The free market alternative for addressing frictional costs is competitive new entry. Helpfully, there is regulatory support and encouragement for this.

MiFID sets a framework enabling and encouraging competitive new entry

There are three areas ripe for competitive new entry in Europe enabled by MiFID

  1. Trade reporting and market data
  2. Trade execution
  3. Cash equities clearing by CCP

Trade reporting and market data – Project Boat

Post-MiFID, there are more transparency obligations and therefore more market data. For example, pre-MiFID, approximately half of German cash equities by value traded away from the Xetra orderbook as over-the-counter (OTC) and legitimately reported nowhere. Deutsche Boerse Group was able to determine this via analysis of its German equities settlement volumes in Clearstream. MiFID mandates a harmonised transparency regime across Europe and thus obliges firms to report previously undisclosed trades, including this previously invisible, but significant, German OTC data. The consequence is a surge of new market data.

Firms, however, are no longer obliged to report only to exchanges. As long as the data format is easily accessible to other market participants and available on a reasonable commercial basis, then MiFID says firms may direct reports to a choice of destinations, including exchanges and other destinations such as a third party or proprietary arrangements. The industry is already responding to this MiFID-enabled opportunity in a way that mitigates the risk of data fragmentation.

On 19 September 2006, nine firms, including UBS Investment Bank, announced Project BOAT, a cash equities MiFID-compliant centralised publication venue for trade reports and systematic internaliser (SI) quotes. BOAT announced its intention to be inclusive and welcome contributions from other market participants. On 22 January 2007, the BOAT consortium announced its selection of technology (Cinnober) and business (Markit) partners. The intended consequences were improved efficiency, reduced reporting fees and pro-competitive challenge to the economic paradigm of market data.

BOAT launched on-time for the 1 November 2007 implementation date of MiFID and within a few months could claim more than twenty firms contributing reports, representing approximately a fifth of European total value traded (i.e. EuroStoxx 600 names as measured by Reuters Market Share Service, January 2008). This is more than that of the largest European exchange, currently the combined London Stock Exchange+Borsa Italiana Group.

In January 2008, Markit purchased BOAT from the banks, and BOAT continues to grow as a first-choice reporting destination. MTFs, like NYFIX Euro-Millennium, are in 2008 choosing to report trades processed on their platforms to BOAT, and this trend of trading venues voluntarily reporting to BOAT has the potential to increase.

Trade execution – pan-European MTF/Project Turquoise/chi-x

MiFID removes domestic exchange concentration rules and recognises three trading destinations: regulated markets (RMs), multilateral trading facilities (MTFs), and systematic internalisers (SIs). Everything else is over-the-counter (OTC).

RMs and MTFs are similar in that both require market surveillance. RMs and MTFs have non-discretionary rules and bring together multiple third party buyers and sellers. Beyond an MTF, RMs verify that issuers comply with disclosure obligations. MTFs can admit to trading a stock name without issuer consent. SIs are firms dealing on own account that also are executing on an organised, frequent and systematic basis outside an RM or MTF. SIs have SI-specific market-wide transparency rules and some protection for firms operating as SIs by limiting access to the firm’s capital to that firm’s clients.

MiFID thus sets a framework for competitive trade execution. The industry is already responding to this MiFID-enabled opportunity.

With internalisation, MiFID encourages the competitive behaviour of queue-jumping, which can enable immediacy of execution, by allowing the ability to ‘trade-through’ a similar price elsewhere historically protected by ‘price-time’ priority. In February 2007, Euronext announced its intention to offer internalisation within its infrastructure and effectively recognise a shift to ‘price-member-time’ (PMT) priority for this functionality.

Interestingly, MiFID and the regulators anticipate that MTFs will be set up by RMs or firms. Already, there have been multiple announcements of intentions to offer alternative trading destinations. For example, a firm, Instinet, announced the MTF chi-x, and an exchange, EASDAQ, announced its plan to launch Equiduct (which may become an RM).

On 15 November 2006, seven firms, including UBS Investment Bank, announced the intention to create a pan-European equities trading platform, Project Turquoise, followed by a statement on 18 April 2007 of the choice of clearing and settlement provider (EuroCCP, a subsidiary of the American DTCC). Aligned with the spirit and letter of the impending MiFID regulations promoting competition, the driver for this MTF is to reduce frictional costs of trading (i.e. exchange orderbook fees) and potentially to innovate (e.g. with some smart anonymous block auctioning to minimise trading impact). This functionality may be analogous to that of BIDS, which stands for Block Interest Discovery Service, another consortium announced on 27 September 2006 by six banks including UBS Investment Bank to establish pro-competitive block trading in America (joined on 1 March 2007 by another six financial firms). On 30 October 2007, BIDS and NYSE Euronext announced their intention to form a joint venture.

Similar to BOAT, the pan-European MTF, today established as separate a company known as Turquoise, plans to be inclusive and open to all qualifying participants that wish to be members. With independent management, Turquoise will operate separately from the banks, and there is no intention to force client flow onto the platform. Turquoise announced September 2008 as its target launch date.

Turquoise will therefore be just one more parallel venue, so the success of this MTF will depend on the attractiveness of its own fees and functionality. Helpfully, key regulators have expressed views supporting this pro-competitive initiative by the banks.

Another MTF, chi-x, has seen matching by international firms of pan-European cash equities grow from EUR0.1bn in March 2007 to more than EUR20bn in January 2008. This is shown in Figure 3 and sets a precedent in Europe for on-orderbook turnover that is not executed on the incumbent where the on-orderbook activity is consistent and growing.

Figure 3: chi-x volume

Source: chi-x

A press statement dated 10 January 2008 announced that a number of firms, including UBS, were taking a minority interest in chi-x. The chi-x market model supports state-of-the-art features including a CCP and full anonymity with respect to broker identifiers. As a live transparent orderbook, chi-x is faster and cheaper than incumbent exchanges. It is worth noting that chi-x has not had a single buy-in since inception, and has already succeeded in processing around a fifth of the global on-orderbook turnover in some individual liquid names. The percentage captured is calculated as the on-orderbook value traded on chi-x divided by the sum of the on-orderbook values traded on chi-x and the incumbent exchange.

Consider Figure 4 which shows that on an example day, 28 February 2008, significant capture by chi-x of on-orderbook activity for key UK cash equities names, including 26% of Prudential (PRU), 24% of Royal & Sun Alliance Insurance Group (RSA), 21% of Legal and General Group (LGEN), 18% of WPP Group, and 17% of Old Mutual.

Figure 4: Percentage of on-orderbook activity in UK stocks captured, 28 February 2008

Source: chi-x

On the same single day, chi-x in total processed more than EUR1.2bn including material capture in other European names as listed in Table 2.

Table 2: Percentage of on-orderbook activity in European stocks captured by chi-x, 28 February 2008

Turnover EUR 1,212,891,150 
Indices AEX 25  6%, DAX 30  3%, FTSE 100  8%, CAC 40  3%, SMI20  1%
Netherlands Hagemeyer 13%, Aegon 12%, Unilever  9%, Ahold Kon 8% Royal Dutch Shell A  8%  
Germany Commerzbank 12,,Adidas Salomon  6%, Metro 6%, Munich Re 6% Deutsche Bank 6%
France Credit Agricole 13%, Danone 6%, Total 5%, Veolia Environ 5%, Renault 4%  
Switzerland UBS  1%, CS Group  1%, Clariant  1%, Julius Baer  1%, Adecco 1%

Source: chi-x

Cash equities clearing by CCP

MiFID access provisions regarding CCP, clearing and settlement arrangements, complemented by the European Code of Conduct signed on 7 November 2006 by European exchanges, CCPs, and Settlement entities, encourage competitive new entry and provide the regulatory tools for users to escalate concerns to European regulators for ‘adult supervision’ if an incumbent attempts to frustrate pro-competitive new entry. The industry is already responding to this MiFID-enabled opportunity.

In consultation with users including UBS Investment Bank, LSE and SIS x-clear announced on 24 May 2006 the intention to provide member firms with a choice of clearing provider in addition to LCH.Clearnet for UK equity trades processed by LSE, from the latter part of 2007. This coincided with publication of a paper by the EU Commission Competition DG entitled, ‘Competition in EU securities trading and post-trading’ which stated that: ‘CCP services could – and probably should – operate in a competitive environment provided issues of interoperability are overcome.’

For some years, a working interoperability precedent has existed facilitating user choice of CCP for Swiss blue-chips as part of the SWX virt-x post-trade market model. Like most offerings by SWX virt-x, this was a result of consultation with users. Interestingly, users, including the platform-neutral pro-competition UBS Investment Bank, suggested ahead of the original SWX virt-x CCP launch that SIS x-clear should not be the only CCP for SWX virt-x as, LCH was such an important service provider to the international markets. The record shows subsequently that SIS x-clear, operating in a competitive environment, won additional clearing business (including that moved to SIS x-clear by UBS Investment Bank) on objective merit via its compelling commercial and functional offerings. Such achievement sends a strong signal that competition works.

It is worth noting that in 2003 users, including UBS Investment Bank, again supported LCH when faced with the prospect of the London Stock Exchange unilaterally redirecting 100% of UK clearing from LCH to EurexClearing. Following user engagement through LIBA, the exchange abandoned its plan. The majority of users, including UBS Investment Bank, then supported the merger of LCH and Clearnet to form LCH.Clearnet on 22 December 2003. Rationale suggested larger scale leading to internal efficiency savings should ultimately deliver lower user fees and thus secure LCH.Clearnet as a ‘magnet’ CCP for incremental flow. Years later, it required the advent of competitive new entry to realise lower fees, particularly for the clearing of cash equities.

Strategic positioning of competitive new entry involves multi-party negotiation and can take time. One initial delay to the launch of UK competitive clearing was due to pre-MiFID test resource constraints by Euroclear UK & Ireland, the settlement entity formerly known as CREST, and resolved for 28 January 2008. Euroclear, listening to users on its UK Market Advisory Committee, to its credit delivered a UK fee cut, in part as compensation for the delay as well as reflecting the strong growth in UK volumes.

UBS Investment Bank publicly voiced support of the platform-neutral pro-competition invitation from London Stock Exchange to SIS x-clear to compete for the clearing of UK cash equities at the time of the original 24 May 2006 announcement, particularly as the Swiss market and SIS x-clear had already opened to LCH. Following successful relevant technical tests with London Stock Exchange, SIS x-clear, Euroclear UK and Ireland and LCH.Clearnet, UBS Investment Bank was the first firm to announce its internal readiness as of 15 February 2008 to take advantage of competitive UK clearing, subject to 24 hours notification of a launch date.

On 10 March 2008, LCH.Clearnet issued a press release stating that, ‘in accordance with the Access and Interoperability provisions of the European Code of Conduct, it will cooperate with SIS x-clear on a peer-to-peer basis in the clearing of trades executed on the London Stock Exchange’ with a launch date soon to follow. Nevertheless, while the implementation of a new market model can take time, during the period since the 2006 announcement of the intention to introduce UK competitive clearing, LCH.Clearnet did announce and deliver multiple pre-emptive fee cuts.

Extension of initiatives such as SIS x-clear to the UK market will give further credibility to rolling out similar competitive initiatives to other European markets that have incumbent CCPs. The elegance of the proposed competitive CCP model open to all is that only members that believe they will benefit commercially and functionally from migrating clearing from the incumbent need switch. The rest stay with the incumbent if they wish. With the pre-emptive fee cuts in anticipation of competitive choice, all users, whether switching or remaining with the incumbent, have realised some benefit. (Users can also choose between LCH.Clearnet and CC&G in Italy for fixed income trades matched on the MTS platform.)

The Code of Conduct exists today and is preferable to a formal Clearing and Settlement Directive that can take years to formulate, by which time commercial opportunities are missed. The multi-year process to agree and implement a Directive risks shaping a legal construct that may be different to original intentions. Therefore, credit is due to the European Commission, the exchanges, post-trade entities, and trade associations like the Federation of European Stock Exchanges, the European Association of Clearing Houses, and the European Central Securities Depositaries Association for the leadership to agree a voluntary Code of Conduct rather than a Directive. Users were not involved in the negotiation of the Code, however. Where the Code of Conduct enables various CCPs to apply for access and interoperability, it makes sense to have some user endorsement of a business case as a filter to avoid unnecessary connectivity for the sake of interoperability.

With competitive new entry comes more choice and therefore complexity

Competition will reduce exchange and CCP fees leading to increased liquidity. Competitive new entry also means more entities and thus more fragmentation. Brokers will need more technology to manage this new complexity, and this will require significant technological investment. Two innovations likely to arrive in Europe will include intelligent or Smart-Order-Routing (SOR) and dark pools of liquidity.

Smart-Order-Routing (SOR)

Many will recall previous attempts at competitive new entry for orderbook trading, including pan-European initiatives from Tradepoint, Jiway, Easdaq, SWX virt-x, Nasdaq Europe, Borsa Italiana’s MTA International, and the more targeted challenges vs Euronext of Deutsche Börse’s Dutch initiative and LSE’s Eurosets Dutch Trading Service. Aside from SWX successfully growing its global market share of its core Swiss blue chip trading by moving its liquidity pool to its London-based SWX virt-x, all previous orderbook attempts to compete with incumbent platforms have yet to gain meaningful market share. This is due to neither lack of good ideas nor lack of resources. One key missing structural component is the lack of SOR mass deployment in Europe.

Because of the overriding importance to participants of liquidity, the current market structure encourages exchange members to continue to direct ‘at-market’ orders (which take offers and hit bids on electronic orderbooks) to the domestic pool of liquidity, and ‘limit’ orders (the actual bids and offers that fill the orderbook) to where those limit orders are most likely to be 'hit' by at-market orders (i.e. again, the domestic market). Unless users have comfort that other market participants can both recognise and seamlessly interact with limit orders placed on alternative platforms, the status quo will remain.

SOR pools puddles. Via intelligent electronic links to multiple platforms, SOR enables seamless recognition and interaction with orders across these physically fragmented platforms as if they belong to one virtual pool of liquidity. Like the fax machine, internet and mobile phone, SOR becomes meaningful with mass deployment.

Why is meaningful deployment of SOR in Europe more probable now than before? Firstly, MiFID as a regulatory imperative meant firms all had to prepare for the same start date, 1 November 2007. All are, legitimately, looking at the same issues at the same time and all are, in parallel, upgrading systems as relevant. Secondly, SOR is available, and many are familiar with SOR methods that have existed within the USA market structure for years. In fact, brokers not deploying SOR in Europe may find themselves at a competitive disadvantage.

How might the incumbents respond to these competitive challenges?

  1. Downplay potential competition and maintain that the status quo will continue.
  2. Slow progress of a competitive new entrant by initiating a regulatory complaint with the resulting bureaucratic process. Such a tactic seems less likely given the awareness and positive comments in the public domain from key regulators supporting the announced pro-competitive initiatives.
  3. Pro-actively reduce fees. This is starting to happen, welcome but small in scale.
  4. Most interestingly, incumbents may leverage their existing offering or announce new competitive offerings. Mass deployment of SOR will increase the success probability of pro-competitive MTFs like chi-x and Turquoise, as well as that of the earlier exchange initiatives listed above. One way or another, competition will reduce exchange fees.
  5. Volunteer to participate in market initiatives. There are also some examples of this.

Dark pools of liquidity

Where SOR pools puddles, dark pools aim to reduce information leakage while finding anonymous liquidity. Dark pools, an increasingly used buzz word, are where orders reside and are not yet executed nor displayed to the market. This is hardly novel. Consider the hidden components of iceberg orders where one has 100 to trade, shows 10 and hides 90. This iceberg order type is classic functionality that exists on many exchanges. Similarly, the traditional matching by brokers of buy and sell orders on the way to the market is another example of accessing dark liquidity.

Dark pools effectively augment traditional broker skills of finding the other side of a trade and automate the process with electronic pipes. More recent examples include alternative third-party systems, such as Liquidnet, BIDS and Pipeline. There are also likely to be improved broker blind crossing of institutional flow and broker blind crossing of institutional with retail flow. Dark pool aggregators like NYFIX Euro-Millennium are similarly launching in Europe.

The etymology of the term ‘dark’ as applied to trading liquidity

Anecdotally, the term ‘dark’ appeared on 23 September 2002, the day the Island ECN decided it would stop displaying key ETF orderbook limit prices as a way to avoid connecting to a relatively slow ITS Intermarket Trading System while still complying with SEC Regulations. Under SEC Regulation ATS in the USA:

‘if an alternative trading system executes more than five percent of the total volume in a given security in at least four of the past six months, it must do two things. First, if the alternative trading system displays quotes, it must display those quotes in the national market system (the Consolidated Quote System for stocks listed on the NYSE or AMEX). Second, it must allow participants at other venues to trade with these quotes. For listed stocks, this means that the alternative trading system must join the Intermarket Trading System (ITS).’

From Island’s perspective, connecting to the ITS would have retarded its relative speed advantage compared with the relatively slow human floor process for trading on AMEX. Rather than join ITS, Island went ‘dark’ and limit orders were no longer visible to any market participant. As Hendershott and Jones described the situation in the Revew of Financial Studies (Vol. 18, No. 3):

‘If an electronic trading system does not display quotes to subscribers at all, it is not required to post quotes or send and receive orders through ITS. Rather than post its quotes and participate in ITS, Island chose to go dark. It stopped displaying its limit orderbook completely in the affected ETFs. It did this by converting all affected orders to undisplayed status…Island explained, ‘We do not believe it would be possible for Island simultaneously to comply with these requirements and to maintain the system performance our subscribers expect.’

There is more than one type of dark pool

Today there is a universe of actionable liquidity that exists in the dark, and one has to be in it to win it. The benefits of using dark pools are minimising market impact by increasing the likelihood of achieving a target price through interacting with anonymous ‘natural’ liquidity while maintaining confidentiality by minimising information leakage. On the other hand, increasing choice means more complexity and, for some, increased difficulty of finding liquidity. A concern is negative selection or ‘pool toxicity’ where stat-arb ‘alpha-hunting’ flow may disadvantage other investor order flow seeking simply to maximise the probability of finding the other side of an otherwise neutral order.

Success at managing an increasingly complex market landscape including dark pools requires smart logic, systems to deploy that smart logic, and a scale of business that can deliver statistically better results while affording to exclude the ‘toxic’ flow. The quest is to seek anonymous liquidity, minimise information leakage, and avoid negative selection.

Pool approaches to handling dark liquidity can be summarised as one of four: scheduled crosses or call markets as within exchanges; alternative crossing systems or negotiation as with Liquidnet or BIDS; external continuous blind crossing like Pipeline; and broker internal blind crossing of institutional or with retail as is routinely done by firms like UBS.

Dark pool: call market example complementing a transparent exchange orderbook

Consider Figure 5, an example of an Xetra orderbook display for RWE, the German power company. The bid prices are on the left and offers on the right. The top of book shows 1884 shares with a best bid of EUR78.61 and a best offer of EUR78.62 for 10 shares. The total shares on the bid are approximately 12,000, and the total shares on offer are approximately 7,000.

Figure 5: RWE orderbook snapshot

Source: UBS.

RWE, at the time of monitoring this example, is a stock that trades approximately 2m shares per day. In the last five minutes on this particular day, 271,000 shares traded at the closing auction. This is more than 10% of the total day’s volume.

One conclusion is that clients are not posting all liquidity on the Exchange orderbook during continuous trading. This relative proportion of orderflow in auction is well known and an example of ‘dark’ liquidity. So, if one wants to participate in this flow, one has to be in it (via the auction) to win it.

This example demonstrates that while continuous limit orderbooks are efficient as a market for ‘price’ and matching small orders, the dark pools are more a market for ‘size’. This behaviour is not limited to Europe. Consider Figure 6 which shows a similar pattern for an Asian market like Korea.

Figure 6: Proportion of daily volume in KOSPI stocks traded at 5 minute intervals

Source: UBS

Dark pool: negotiation example

In negotiation models like that of Liquidnet, buyers and sellers are alerted if a potential matching partner is interested in a particular name. Engaging the alert for potential negotiation initiates an anonymous chat window within which dialogue can lead to an execution of agreed price and size followed by a trade report to the market only after the trade. Some negotiation models have a feature or a star-like ranking indicating the quality of the anonymous counterparty, like an e-Bay rating. Here this rating feature provides two functions.

First, it allows one to determine whether the anonymous counterparty is likely to be serious (a high rating) or just fishing for information and unlikely to trade (a low rating). Secondly, the ranking system conveys an element of self-regulation or encourages a form of self-control as one knows that by entering into such an anonymous negotiation, the result of the dialogue will impact one’s own rating, i.e. not trading may lead to a lower rank anonymously displayed ahead of the next potential negotiation which may impact a future counterpart’s willingness to trade.

Dark pool: continuous blind example

A continuous blind system is designed to operate like a continuous matching system while specifically minimising information disclosure that might otherwise signal direction of the firm orders in the system. Figure 7 shows an example of a Pipeline screen. Orders are entered into Pipeline and the only indication that interest exists is that the symbol changes colour to orange. The market is not told whether the interest is buying or selling or at what price. If another firm order enters and interacts with an existing order, then the computer system matches the two orders and informs the market through post-trade reports. Order types facilitate the potential for mid-point matching that may lead to price improvement. Otherwise nothing happens. The confidentiality of the intentions of the residing orders is preserved.

Figure 7: Pipeline’s continuous blind dark pool


In the United States where exchanges and trading platforms operate in an environment of competition, less than half of a share’s turnover actually trades on the market on which it is listed, as shown clearly in Figure 8. Significant flow is executed away from traditional exchanges and reported to the Trade Reporting Facility (TRF).

Figure 8: US market share (26-29 March 2007)

Source: UBS estimates by Matthew Scoble

FINRA is the Financial Industry Regulatory Authority, the largest non-governmental regulator for all securities firms doing business in the United States. FINRA explains that each FINRA Trade Reporting Facility (TRF) provides FINRA members with a mechanism for the reporting of transactions effected otherwise than on an exchange. While each FINRA TRF is affiliated with a registered national securities exchange, each FINRA TRF is a FINRA facility and is subject to FINRA's registration as a national securities association. Trades by FINRA members in Nasdaq-listed and other exchange-listed securities, as approved by the Securities and Exchange Commission (SEC), executed otherwise than on an exchange may be reported to FINRA TRF.

When one investigates further the relative contributions to the TRF reported flow, one expects contributions from the various dark pools that exist in the United States. What is interesting about the TRF breakdown by volume as shown in Figure 9 are two observations. First, each of the key third-party dark pools, like Liquidnet and Pipeline, represent individually a small slice of the executed flow. Secondly, the aggregate executions processed by the third-party dark pools still represent a minority of the analysed flow.

The overwhelming observation is that the largest contributor to the processing of dark liquidity in the market remains internalisation, and this means that brokers represent the largest class of dark pool.

Figure 9: TRF breakdown – daily volume (m)

Source: UBS estimates by Matthew Scoble


In the past, a debate focused on whether internalisation negatively impacts the price formation process. What do we mean by internalisation? Internalisation is simply order-routing, crossing, and market-making. This behaviour is well-known and established as market practice. Automation speeds and improves the process.

Internalisation is about competition and providing more liquidity, better prices and faster execution for clients. One consequence of internalisation is the pro-competitive tool of queue-jumping when providing immediacy for clients.

Consider the following example. Where a client wishes to sell stock and the market shows a bid of 3, a broker preferred by the client may match the 3 to fill the client with the consequence that the client has queue-jumped a firm waiting earlier in the market.

Systematic internalisation is enshrined in MiFID, and internalisation has been a feature of the USA market structure for years, particularly via the Nasdaq model. In the past there had been a challenge by particularly the NYSE floor, but empirical observations demonstrate that internalisation, like that within the Nasdaq model, does not disadvantage participants compared with the NYSE floor model. Figure 10 shows that the price formation process as reflected in effective spreads for names via the Nasdaq pro-internalisation model are as good as if not better that those via NYSE.

Figure 10: Internalisation vs central limit orderbook

Source Nasdaq

In fact, NYSE Euronext seems also to have endorsed internalisation with the 2007 introduction in Europe of its Internal Matching Service supporting PMT ‘price-member-time’ priority for this functionality, as described above.

Dark pool: broker example including internalisation

Brokers are continuing the traditional skill of finding the other side of the trade and, through automation and system-embedded checks, safely speed the process via electronic pipes. UBS is one of the largest liquidity pools in the world, and the UBS internal liquidity pool is leveraged in the United States as the UBS Liquidity Network. Figure 11 shows how the inclusive UBS model is able to perform a fair broker-blind crossing of institutional flow and broker-blind crossing of institutional with retail flows.

Figure 11: UBS Liquidity Network, example United States

Using the UBS Price Improvement Network (UBS-PIN), orders are not obliged to be reflected in other venues or exchanges, and anonymity eliminates costly information leakage. UBS direct execution algorithms have access to meaningful liquidity pools and puddles. UBS is able to provide anonymous access to retail flow and the option to interact with institutional flow. Advantages UBS offers that are key differentiators include the scale and quality of the UBS liquidity pool. As the most liquid dark pool, there is the statistically increased probability of a fair result with potential price improvement via crossing. And because of the scale of the UBS liquidity network, UBS can afford to maintain a positive diversity of orders and exclude ‘toxic’ flow and therefore minimise negative selection. The resulting pool of retail flow complements that of institutional flow, and the steady nature of the flow leads to a lower risk of a large execution at the wrong price or time. Individual UBS-PIN executions can be small, but overall the volume processed is large.

UBS has offered anonymous interaction with retail flow for some time as part of UBS-PIN. It is a fact of ranking tables that UBS has the largest pool of retail flow on Wall Street, and this is a differentiator for UBS. The UBS Liquidity Network is a natural extension of existing business and is part of the service offered also to UBS wholesale customers.

The emergence of dark pools in Europe

The TABB Group counts at least 40 crossing networks and dark pools in the United States and expects that ‘trends point toward increasing market structure fragmentation and the spreading of electronic trading tools across geographies.’

With NYFIX Euro-Millennium launching its European service in March 2008 and joining ITG and Liquidnet, Europe will have three operational third-party dark pools. Additional platforms are expected to follow given the web presence of entities like OpenMatch, announcements like the proposed joint venture by NYSE Euronext with banks in Europe for Project SmartPool, the expected European arrival of Pipeline, and the introduction of volume discovery services and ‘dark’ order functionality of existing incumbents like Deutsche Börse and new entrants like Turquoise. SWX Europe announced that it would deliver its dark pool functionality by exclusively white-labelling NYFIX Euro-Millennium for Swiss flow and process this via its choice of existing and evolving post-trade solutions. Of course, broker liquidity pools already exist in Europe. A key insight to draw is that the dramatic growth in European exchange on-orderbook activity, as in Table 1 above, shows clearly that a market structure embracing the ability of brokers to internalise or execute off-orderbook or avail themselves of dark pools can complement a positive growth in overall liquidity.

The challenge to manage more complexity

The challenge is to source liquidity in an increasingly fragmented landscape. This is therefore not only about technology, it is also about improving process.

MiFID itself provides an example with Best Execution. Best Execution in many jurisdictions today emphasises price. MiFID redefines Best Execution as a ‘Process to deliver Best Possible Result.’

Under MiFID, firms (both buy-side and sell-side) have new Best Execution obligations which are broadly to have an execution policy to take all reasonable steps to achieve the best possible result for their respective clients and to be able to demonstrate on request from clients or regulators they have executed orders in accordance with their policies.

Research is no longer a criterion for choice of executing broker. Building on the trends of unbundling legislation adopted since 2006 in the UK, buy-side clients have freedom to direct orders to the destination that gives Best Execution. Separately, buy-side clients have the power to reward value-added research, for example, via Commission Sharing Agreements (CSAs). The result will be more competition among firms.

What skills will clients increasingly demand? These will include crossing, pro-active liquidity finding, and competence deploying quality technology.

On what criteria will brokers seek to differentiate themselves? Market share and the quality of internal liquidity access will be critical to a broker’s crossing performance. The logic for buy-side clients will be to direct order flow to the brokers with larger market share and better internal liquidity, since this will increase the probability of crossing and therefore the probability of potential price improvement leading to better investment performance.

Connecting only to an exchange’s orderbook will miss all the potential dark liquidity of the leading broker. Pro-active liquidity finding is all about the traditional brokerage ability of confidentially finding the other side of the trade. Confidentiality and minimising information leakage will highlight the increasingly important need to interact with a broker that stands by a policy in public of, for example, ‘No pre-hedging ahead of client orders.’

Technology competence will include connection of Smart-Order-Routing to a meaningful number and range of multiple venues, algorithmic trading for minimising market impact on deployment of order execution strategies, and a structured process for monitoring and evaluation.

Change is a dynamic process

Buy-side clients embracing change can have more importance if they more effectively exercise their responsibilities to understand, explain, monitor, decide, and justify their choice of executing broker. This may require some buy-side clients to make new efforts to learn about the state-of-the-art services of their sell-side execution brokers, including, for example, how algorithms work and how directing orderflow to the best executing brokers help the buy-side client better compete with other buy-side peers. The process to deliver best possible result by brokers will increasingly extend to include sales/trading complementing highest consistent execution quality with calls of relevance and insight.

The impact of MiFID

MiFID creates opportunities for those delivering the highest standards and duty of care.

MiFID will increase competition, lower exchange and CCP fees, and increase liquidity. MiFID will increase choice and therefore complexity, and Europe will see Smart Order Routing deployed. MiFID will increase change adding importance to those buy-side clients building on unbundling trends and exercising their new responsibilities to understand, explain, monitor, decide, and justify their choice of executing broker on MiFID criteria that excludes research, while separately rewarding value-added research.

Ultimately, MiFID sets a framework where orderflow and new business can concentrate to those most capable in an increasingly fierce commercial and technical arms race.