Navigating Liquidity 4

Transcription

1 EUROPE April 2010 Navigating Liquidity 4 Market microstructure: a paradigm shift Pre-fragmentation, market participants' roles could be clearly defined as investors, intermediaries or market operators. Interactions with markets were simply labelled as either liquidity providing or liquidity consuming. Reg NMS and MiFID caused a paradigm shift in this structure by radically changing the roles, interactions and microstructure of the market. Now that regulators are discussing possible adjustments to Reg NMS and MiFID to improve the efficiency of the "market of markets", it is time to fully understand how the new microstructure has changed the various roles and interactions. Thus, the frequently-asked questions can be restated according to this paradigm shift: Pre- and post-trade transparencies? Post-MiFID, the pre-trade is blurred by dark liquidity and high frequency orders, so how fast and reliable should the post-trade data be to provide enough information to the market to make efficient decisions? In our view, we must be ready to trade in a probabilistic universe rather than a deterministic one! High Frequency Trading threat? Since HF traders act more as market operators than as investors with a long-term view on the price, is there a frequency bound above which they will generate more "market impact amplification" than stability to the price formation process? As we provide evidence of the existence of such a threshold, the question becomes: has it been crossed, and if so, what can be done? Dark pools contribution to the price formation process? Considering dark pools as electronically reachable OTC pools, what is the nature of the liquidity flow they provide to the market? With the benefit of CA Cheuvreux's executions in the dark, we shed light on the nature of this dark liquidity and its link with the bid-ask spread on the primary markets. Best execution? The balance between immediate price improvement, execution speed and stealth has to be solved for each investor according to his risk/reward profile. We present a liquidity capturing scheme, implemented in our global CrossFire algo, that suits the specificities of the actual liquidity landscape. In this report we also provide a list of the 3 main questions the buy-side, the sell-side and corporates should ask their brokers about the market microstructure. Disclosures available on

2 Below, we articulate the three most important questions that each of the various participants should ask about the market microstructure. Buy-side institutions What is the intra-day impact on the liquidity risk of a stock? During the allocation process, a part of the liquidity risk comes from intra-day (i.e. from the microstructure). Proxies like the bid-ask spread have to be weighted by the liquidity available at the touch. Is the breakdown of my executions on trading venues in line with my risk aversion? Being passive during an execution is in line with low risk aversion, while being aggressive is required when there is a highly risk-averse profile. Do I need to execute my trades in the dark? From a buy-side point of view, dark pools can be considered OTC pools. They are useful for large orders (i.e. a multiple of the average trade size or "large in scale" orders) if they can provide anonymity or execution speed. Sell-side institutions Is my latency to each market lower than its Nyquist-Shannon latency bound? If not, you have to consider switching to a non-deterministic mode in order to achieve optimal routing. Do I control the balance between execution speed and adverse selection (especially in dark pools)? Being executed in the dark by an HFT* who immediately unwinds in a visible book provides you with speed and delegates to the HFT* the splitting of your order into small parts. Of course you pay a price for that, but make sure that it is a fair price with respect to your need for speed. Are my liquidity-seeking tactics self-adaptive? The liquidity flow provided by the market changes rapidly and is different from one pool to another. A liquidity seeker should have self-adaptive features, taking into account the variations of the liquidity flow, like CrossFire. Corporates Does the HFT* change the profile of my shareholders? HF trading strategies, especially market making ones, structurally unwind their positions before the close. They therefore have no impact on the profile of the shareholders. What are the specificities of dark pools from my point of view? From a corporate point of view, there is no difference between an OTC deal carried out in a broker's book and one that is executed in a dark pool. What is the turnover of my stock on visible books, dark pools and OTC deals today? The reporting processes are not taking place on the same timescales: as deals on visible books (primary markets and MTFs) are reported "as soon as possible"(i.e. a few milliseconds after the occurrence of the deal) via low latency connections, there is more blurring on OTC deals and dark pool ones. Some dark deals are published using the same channels as visible ones (for instance, deals on "mid-point pools") and others have proven to be duplicated. *HFT: High Frequency Trader 2

3 Introduction With MiFID currently under review, this issue of Navigating Liquidity covers two main topics: an in-depth analysis of the changes in market microstructure, and the mechanism of dark pools. Our overview of the changes in market microstructure aims to be general and uses tick-by-tick data. It includes a comparison of the intra-day volatility of Spanish stocks with that of UK stocks. We discuss all the important points that are currently under discussion: high-frequency traders, the role of the players, the positioning of investors relative to the system, etc.. We also present our new CA Cheuvreux fragmentation index (or CFI); this will be included in our Monthly Liquidity Indicators to monitor the European fragmentation. Thanks to our in-house tick-by-tick database, we present an in-depth study of the characteristics of dark liquidity. Lastly, we describe a liquidity-capture scheme, which is the basis of our main liquidity-seeking algorithm: CrossFire. About the authors Charles-Albert Lehalle Currently Head of Quantitative Research at CA Cheuvreux, Charles-Albert Lehalle also lectures at "Paris 6 (EL Karoui) Master of Finance" (Ecole Polytechnique, ESSEC, Ecole Normale Supérieure) and MASEF/ENSAE one and gives master classes in the Certificate in Quantitative Finance in London. He has also given lectures at numerous seminars and international conferences at MIT, Ecole polytechnique, Columbia, NYU, etc.. With a Ph.D. in applied mathematics, Charles-Albert is an expert in stochastic processes, information theory and nonlinear control. He has published international papers on quantitative finance, real-time optimisation of high dimensional processes (with applications to Formula One, high-mix fabs, large plants, and aerospace) and learning theory. Romain Burgot Currently working as a statistician researcher in the CA Cheuvreux Quantitative Research team, Romain previously worked on microstructure questions for the final paper of his studies (ENSAE 2006), which was entitled "The comparison of VaR measures with high frequency data". Execution Services Contacts GENERAL HOTLINES: Paris: London: New York: Ian Peacock Global Head of Execution Services / ipeacock@cheuvreux.com SELL SIDE BUY SIDE Jonathan Carp Mark Freeman Head of Alternative Execution Sales Europe Head of Alternative Execution Sales to Buy Side / jcarp@cheuvreux.com / mfreeman@cheuvreux.com 3

4 CONTENTS Introduction 3 I A new global microstructure: towards a paradigm shift 5 Changes in the microstructure 5 A blurring of roles 11 Impact on liquidity 12 The Spanish experiment 14 II A new way to describe the liquidity discrepancy 17 Entropy of the market microstructure 17 An entropy-based index: the CFI 17 III Dark pools, an alternative liquidity source 20 Mechanism of dark liquidity pools 20 In-depth analysis of dark liquidity 21 IV An optimal liquidity capturing scheme 27 A glimpse at CrossFire s tactics 27 CrossFire efficiency 28 Appendices 30 Appendix 1: Glossary 30 Appendix 2: Lists of stocks

5 I A new global microstructure: towards a paradigm shift Changes in the microstructure The microstructure of a market is defined as the way the different traders share their responsibilities and their roles. The rules are defined by the regulators and under these rules (mainly respecting regulatory-defined constraints), each trader works to maximise his utility (i.e. a risk vs. reward balance). In the European and US equity markets, the traders are usually divided into two main categories (see Trading & Exchanges, by Larry Harris, 2003): The agents (brokers and buy-side traders); The principals (utilitarian traders, speculators, dealers and "futile" traders). As underlined in previous issues of Navigating Liquidity, the characteristics of liquidity available on a given stock is a mix of the trading rules, the type of traders trading it, and the properties of the underlying asset (i.e. the listed firm: its sector and its financial health). MiFID, which came into effect on 1 November 2007, changed most of the rules for the equity market through the introduction of two main guidelines: The liberalisation of trading venues: by abolishing the concentration rules in place in some European countries, MiFID enabled the creation of "alternative" trading venues. The concept of "best execution": this is not clearly defined in the regulatory framework, but each broker is allowed to draw up its own definition, and thus provide access to the market through a clearly defined "best execution policy". FIGURE 1: DIAGRAM OF THE MARKET STRUCTURE BEFORE AND AFTER MiFID Before MiFID Today Investors HFT Investors OTC Intermediaries SOR Crossing engine Dark Pools OTC Intermediaries Visible book Dark pool Mid point Visible book Dark pool Mid point Hub Visible book Primary MTF MTF MTF MTF Primary 5

6 The liberalisation of trading venues first spread the liquidity among a few venues, each of which offering several pools Brokers and buy-side traders inherited the crucial role of consolidating the liquidity Some of the outcomes of these changes were unexpected. However, the liberalisation of the trading venues did not directly lead to the creation of numerous alternative venues (the financial crisis clearly could not encourage investors to finance newcomers), but rather each of these venues (Chi-X, Turquoise, BATS Europe, NASDAQ-OMX), now offer a multitude of "liquidity pools" (visible pools, anonymous ones, etc). Most primary markets now also offer alternate liquidity pools. Rather than giving rise to a high level of competition, the liberalisation of trading venues spread the liquidity among a few venues, which rapidly split "their" markets into niches. As most final investors' needs consist of a mix of different liquidity pools, the agents (i.e. brokers and buy-side traders) inherited the crucial role of consolidating the liquidity (Figure 1). It is worth noting that the fragmentation followed a very similar path in the US, even though the "equivalent" regulation (Reg NMS) came into effect two years earlier. The role of the financial crisis, even if it is difficult to isolate, may be a very important common factor to explain the almost simultaneous emergence of such a microstructure in both regions. In addition, the concept of best execution focused on "immediate price improvement", which is mainly dedicated to aggressive orders and which is based on snapshots of visible order books for small orders. The split of larger orders into smaller ones is delegated to brokers via his "execution policy". This was clearly a rational way to define best execution before the fragmentation of the market. However, after more than two years, we think that best execution has not to be considered by investors outside the execution policy of his broker, given, on the one hand, the importance of capturing liquidity passively, and, on the other, the lack of information recorded in snapshots (hidden orders have increased by 30%, according to NYSE Euronext, some agents and high-frequency traders use mirrored orders; invisible posting is possible on visible and "dark" books). It is now more about "optimal execution"; i.e. the conjunction of splitting tactics with immediate price improvement for aggressive orders and more subtle criteria (as opportunity costs) for passive orders resulting from the splits. At present, optimal execution is first and foremost a matter of offering consolidated access to liquidity for both passive and aggressive orders. As is the case for Crédit Agricole Cheuvreux with BLINK, most efficient brokers offer access to an internal crossing engine, in order to reduce information leakage. They also give access to anonymous pools via trading algorithms and Smart Order Routers (SOR). 6

7 All the available liquidity has to be consolidated to meet the specific needs of investors Most regulators are now considering modifying the rules (note the SEC's concept release, the Fleuriot Commission in France, etc.) and it is thus difficult to define exactly the right benchmark at the present time. It nevertheless seems that best execution can no longer be limited to the split of an aggressive order according to the snapshot of several order books. It is about how the available liquidity can be consolidated for the specific needs of each investor or trader. In the realm of market microstructure, the clearing and settlement channels play a major role, as it is meaningless to compare millisecond-time stamped order books of two venues that do not share the same clearing and settlement costs and risks. MiFID did not directly push for a normalisation of these aspects, as has been continuously stressed by most of the players. FIGURE 2: DIAGRAM OF THE DECISION-MAKING FRAMEWORK Market observers Trader 1. Information (pre trade transparency) 2. Send orders 3. Result (deal) as a private information 4. Public post trade transparency Trading venues 7

8 The paradox that market design has to answer is to provide enough information to ensure a fair price formation process, while reducing information leakage and avoiding front-running The development of dark pools and HF trading shook up the principles of MiFID The importance of the market microstructure stems from the fact that we all entrust this market of markets with the price formation process. This process is one of the mechanisms at the heart of capitalism: it should, by balancing supply and demand via the occurrence of deals between traders, form prices that constitute a fair view of the value of the exchanged assets. To guide traders to decide at which prices and for which quantities they have to place orders, the market has to provide them with information prior to the trades and then make public the information that a given trade took place (Figure 2). The paradox is that, on the one hand, the more information that is available to the traders, the better the price formation process (PFP) will theoretically be, but, on the other hand, each trader fears "information leakage" to avoid adverse selection (being front run) or having an excessive price impact. Market design is clearly at the heart of this contradiction: the microstructure should allow enough information to be shared by all investors to ensure a fair PFP, but the interest of each investor has to be protected from information leakage. Reorganising the market microstructure is thus a difficult and subtle task, involving changes in the way information goes from one trader to all other market participants and vice versa. It is clear that any investor will primarily be concerned with protecting his own interest, participating with stealth in the PFP, but some investors are less capable of protecting themselves, especially if this protection involves heavy technological investments. The issue of protection of all investors can only be resolved through regulation, and MiFID aimed at this protection with the concept of pre-trade and post-trade transparency. Demanding all participants (the investors themselves, intermediaries, and trading venues) to publish enough ex ante and ex post information has been a move towards guaranteeing a fair price formation process. While the constraints on pre-trade transparency are clear, there is a consensus about the difficulty of obtaining efficient post-trade transparency. Another principle underlined by MiFID was to use the size of an order as a proxy for the level of maturity of a trader: the larger an order, the more autonomous the investor is; the smaller the order, the more the originator of the order has to be protected. Two developments recently shook up these principles (the sharing of information and the relationship between the size of an order and the maturity of its originator): "dark pools" and high frequency traders (HFT). The former do not publish the same level of information as the other liquidity pools, and the latter are highly informed investors using only very small orders (the increase in the number of trades independently of the decreasing average trade size is often considered as evidence of the increasing activity of HFT, see Figures 3 and 4 below). 8

9 FIGURE 3: INDICATORS OF AVERAGE TRADING SIZE (ATS) AND FIGURE 4: INDICATORS OF AVERAGE TRADING SIZE (ATS) AND DAILY NUMBER OF TRADES FOR THE FTSE 100 DAILY NUMBER OF TRADES FOR THE CAC 40 Indicator of the ATS Indicator of the ATS Indicator of the no of deals /09 06/09 03/09 12/08 09/08 06/08 03/08 12/07 09/07 06/07 03/07 40 Indicator of the no of deals 02/10 11/ /09 03/09 12/08 09/08 06/08 03/08 12/07 09/07 06/07 03/07 02/10 11/09 09/09 As it is very difficult to define HFT clearly, merely recall the definition given in the SEC's concept release (page 45): 1) "The use of extraordinarily high-speed and sophisticated computer programs for generating, routing, and executing orders; 2) Use of co-location services and individual data feeds offered by exchanges and others to minimize network and other types of latencies; 3) Very short time-frames for establishing and liquidating positions; 4) The submission of numerous orders that are cancelled shortly after submission; 5) Ending the trading day in as close to a flat position as possible (that is, not carrying significant, unhedged positions overnight)." Dark pools are also difficult to define, even if their regulatory status is quite clear: in short, they are multilateral trading facilities (MTFs) governed by a waiver allowing them to match orders without pre-trade transparency (when the size of the order is small enough) as long as the price of the deals is imported from a visible reference market. The rationale of such a rule is that from a theoretical point of view, deals at imported prices do not participate in the price formation process. Dark pools and high frequency traders share a common point: they blur the quality of the liquidity available in order books; they thus modify the usual definitions of liquidity providers and liquidity consumers. The trading activity of HFT updates limit order books at a higher rate than the round trip for any non co-located observer. The snapshot of the limit order books you take into account to make decisions does not, in most cases, reflect the state of the book when your order actually reaches it. This can be taken into account using real-time estimates of the quantity that is really available in the books at each price. This is a shift in the paradigm: a few years ago, one could send an order to a market knowing that it would be a liquidity provider or a liquidity consumer. Now, when an order is sent to the market, the present information only gives an estimate of its probability of capturing aggressive or passive flows. 9

10 Such a situation has to occur once the trading frequency crosses a given threshold. This is a well-known result of information theory known as the "Nyquist-Shannon sampling theorem". Restated in trading terms, it says that as soon as the frequency of the changes in the order book of a trading venue is greater than half of the inverse of the latency between an observer of the venue, the observer loses information. Our Nyquist-Shannon trading threshold formula is consequently: 1 1 Order book update frequency threshold = 2 latency This implies that if a trader is at a "distance" (i.e. latency) of 2ms from a trading venue, he has complete information as soon as the order book update frequency is lower than 250Hz. If not, the time for him to have the information and to send back an order is higher than the time to the next update. The Nyquist-Shannon latency bound is now over 2 milliseconds on large caps We will call the Nyquist-Shannon latency bound of an order book the latency needed to fulfil the eponymous threshold: Nyquist Shannonlatency bound 1 = 2 1 Order book update frequency Figure 5 shows the variations of the number of updates of the order book (five limits) per second (i.e. frequency in Hertz) for two liquid stocks (one listed in London, and the other in Paris) during the day. It is greater than 250Hz during most of the day, which implies a Nyquist-Shannon latency bound of 2ms. FIGURE 5: INTRA-DAY FREQUENCY OF UPDATE OF THE FIRST FIVE LIMITS FOR TWO LARGE CAPS IN EUROPE (ON THEIR PRIMARY MARKET AND ON A CONSOLIDATED TAPE): 250HZ IS A COMMON VALUE, ITS ASSOCIATED LATENCY BOUND IS 2MS 250Hz Dark pools also contribute to this blurring of definitions: on sending an order to such a liquidity pool, the only information you have is the probability of being a liquidity provider or liquidity consumer. 10

11 The contribution of a trader to the price formation process is no longer deterministic Dark pools are more compatible with the new trading landscape than is often said This uncertainty about the nature of the contribution of an order to the price formation process is now so high that a new type of order is available (for instance on Xetra and NYSE Arca Europe): "book or cancel". Such an order is cancelled if it happens to consume liquidity. The necessity of such a type of order can be read as acceptance by the markets and by most of the traders that they cannot know definitely the nature of their contribution to the PFP when they send an order. These two points emphasise the importance of post-trade transparency: the fact that a trade occurred at a given price for a given size gives clear deterministic information on the price formation process. The weight of such an information in a decision-making process is far greater than that of a snapshot of an order book: at present, pre-trade transparency carries less importance than post-trade transparency. Unfortunately, the quality of post-trade transparency is lower than for its pre-trade counterpart. Clearly a review of current regulation should improve the quality of the former. As dark pools do not provide pre-trade transparency (they import the price from an external venue because the pre-trade transparency of such a venue is theoretically considered to be a good proxy for the nature of the liquidity they are able to provide), but do provide post-trade transparency, dark pools are more compatible with the new trading landscape than is often said. Dark pools are liquidity pools for which the price formation process takes place anonymously, and their outcomes (deals) are publicly reported, as for "lit" pools. Some investors may feel more comfortable dealing under the cloak of anonymity, in order to minimise the amount of information they give the HF trader. A blurring of roles Not only has the frontier between liquidity providers and liquidity consumers become less clear-cut with the change of microstructure, but the separation of roles between the different operators and intermediaries of the market has become blurred. As emphasised by Figure 1, the previous situation was simple enough. There were: Primary markets with listing and surveillance capabilities; Intermediaries in charge of pairing interests between investors in-house (via OTC trades for instance) or on visible markets. In this simple pre-mifid model, the market share of OTC trades was not clearly known and often underestimated by observers (investors and corporates alike). Now the roles are less clear-cut: MTFs have been created by investment banks (e.g., Turquoise), brokers, overseas exchanges (e.g. BATS Europe and NASDAQ-OMX), or independent shareholders; European primary markets also govern or own MTFs with visible (NYSE-Arca) or anonymous books (e.g. SmartPool or Xetra midpoint); A proportion of previously OTC trades now takes place in the "crossing engines" of brokers (BLINK, for example), some of which are regulated as MTFs; Brokers are struggling to give their clients a consolidated view of the pools, given that each client needs a specific mix of liquidity pools; MTFs offer more than one trading pool (mid-points and dark pools) and also give access to the crossing engines of some brokers (such as TQ Lens); The HF traders, previously considered as investors, are now closer to being part of the price formation process (i.e., their role is closer to that of intermediaries). 11

12 When the positioning and role of a player in the market changes, the regulator should ensure that this will not cause efficiency leakage in the system Competent entities have to be designated to carry out market surveillance The fact that a system has become more complex is not a bad thing in itself. On the contrary, it is often an opportunity to achieve progress. For instance, if you are able to use a plane instead of a car, you will be able to reach more destinations and to follow new trajectories, although flying a plane is more difficult and needs more traffic control than driving a car. However, in one sense, we are in the situation where everybody can now fly a plane despite the fact that infrastructures have not been developed at the same pace (post-transparency) and that the air traffic control cannot adequately monitor the situation given the new micro-structure! In the field of market design, one must bear in mind that each participant will try to optimise its own utility. This is the root of liberalism, popularised by the welfare theorem: in a well-organised market, the fact that each player aims to maximise its own utility will lead to an efficient state for all concerned, without there being a need for dialogue. When the complexity of the market microstructure increases, one must ensure that the relationship between the utility of each player and the efficiency of the whole system is preserved. The change in positioning of some players, such as the high frequency traders (HFT) which simultaneously drift from being investors to market operators and to the leading clients of trading venues, can be considered a danger from this point of view. As was the case in the US last August, exchanges can maximise their utility (i.e. increasing the satisfaction of their clients) by offering "flash orders" to HFT as if they were market operators. As the latter retain the utility function of a regular investor, their use of such information leakage to their own interest will not benefit the efficiency of the market. When the positioning and the role of a player in the market changes, the regulator should ensure that this will not cause efficiency leakage in the system. In addition, competent entities have to be clearly designated to carry out market surveillance. Primary markets claim that they support most of the costs of surveillance. Those costs are clearly shared by brokers, which also have regulatory constraints of this nature. Moreover, the primary market receive revenues generated by market data. As this point is also an issue under discussion, the answer could be for such revenues to be a reward for the entities designated to carry out market surveillance. Impact on liquidity The price formation process (PFP) has long been studied by academics and practitioners. A consensus seems to exist around the idea that the PFP consists of the impact of each trade occurring in the market, combined with the variation of the "fair price" of the listed asset. This "fair price" is very difficult to estimate and the role of a large broker as an intermediary is to provide information about the fair price via financial analysis (i.e. "research") on the one hand and to concentrate a large part of the liquidity (i.e. execution) on the other. It positions such brokers at the intersect of the two main forces driving the PFP. In order to give a simple view of the PFP, we split it into two phases (see Figure 6): The deals impact the price (through the balance between buyers and sellers, and liquidity-providing orders and liquidity-consuming ones) in conjunction with moves of the bid ask spread. This is often named the "immediate price impact" (as presented by F. Lillo, J. D. Farmer, R. Mantegna in issue No. 421 of Nature, Jan. 2003). Liquidity returns (or not, as the case may be) according to the distance between the best bid and ask prices and investors' view of the "fair price". This leads to a "decay" of the impact, also termed the "resilience" of the limit order book. 12

13 FIGURE 6a: PRICE FORMATION PROCESS IN TWO STEPS: IMMEDIATE IMPACT (LEFT) AND THEN DECAY (RIGHT) Immediate impact 103 Decay Price 101 Price Time Time As illustrated in Figure 6a, the price impact of a deal can be absorbed rapidly if the impacted price is far from investors' views of the fair price, generating a temporary market impact, or can move the price permanently if their views of the fair price move in accordance with the new quoted prices. The role of post-trade transparency is underlined here: potential traders have to be warned that the latest price is not in line with their view, so that they can send limit orders to the market (i.e. provide liquidity to it) and bring the price back to a more efficient level. The more investors with views of the fair price are part of the PFP, the more likely the decay after a deal will reduce the price impact rapidly The faster the information is propagated, the sooner the impact will decay thanks to investors having a view of the fair price, or which utility function is consistent with this liquidity providing role. For instance, market markers with inventory-keeping constraints can contribute to the decay of the price impact without having any view of the fair price. As a result, the more investors with views of the fair price are part of the PFP, the more likely the decay after a deal will reduce the price impact rapidly. If the horizon of an investor is longer than a jump in the price, he will probably contribute to the subsequent decay. Conversely, the inventory of a market maker has to be larger than the turnover of a jump in price to allow it to contribute to the decay. 13

14 FIGURE 6b: THE HORIZON OF TREND-FOLLOWING STRATEGIES IS LOWER THAN THAT OF TREND-REVERTING ONES The smaller the horizon of the HFT, the more they will act as "market impact followers" MiFID has not yet been implemented in Spain due to the specificities of its clearing and settlement legislation The growing contribution of HFT to the price formation process can be seen as a threat to the stability resulting from the decay in price impacts. HFT do not necessarily have a view of the fair price of the stock they deal, and high-frequency market makers have small inventory limits. The shorter their horizon is, the less they will be able to provide liquidity to decay the impact of the trades. They will act as "market impact followers" rather than "market impact reverters". This observation questions the possible contribution of the HFT to the stability of markets. The Spanish experiment Theoretically, MiFID appeared in Spain at the same time as in any other European country: on 1 November However, due to the legislative complexity of the Spanish market, MTFs appeared very late on and still have very weak market shares. For example, Chi-X's market share on the IBEX 35 only exceeded 0.5% for the first time in August Its market share is still limited at the beginning of 2010, with a peak at 1.43%. Turquoise and BATS are in the exact same situation, with even weaker market shares. This situation is mainly caused by the legislative difficulty for investors to carry out the clearing and settlement of MTF trades. In fact, the company in charge of clearing and settlement on the Spanish market, Iberclear, belongs to the main marketplace. The Spanish system strives to prevent MTFs from being present in Spain by imposing a register code prior to the clearing of any transaction, which implies high costs. Clearing and settlement on MTFs will really only be possible once Spain has changed its legislation so as to open up the clearing process to competition. The market regulator CNMV has published a text stating that these improvements will occur within approximately two years. Once these movements come into effect, other clearing companies will be able to come to Spain, most probably prepared to provide MTFs with access to clearing. Until that date, and for the past few months already, MTFs have found a way of being present, through BME. At the end of the day they cross transactions involving all their volumes with BME brokers. This implies fees from the BME in addition to the usual fees for clearing and settlement. This makes MTFs much less competitive than they should be, and prevents them from entering into direct competition with the main venue on the market. This explains the very weak volumes of MTFs lately. This situation is expected to improve in the next two years, maybe sooner if the European Commission asks for faster transposition of MiFID to the Spanish market. 14

15 Spain can be used as a control to study the impact of MiFID on intra-day volatility In conclusion, Spain has not been impacted by the expected fragmentation arising from MiFID. This specific feature makes the Spanish market a relevant frame to study the behaviour of other European markets that applied the European directive properly. Thus, comparing the FTSE 100 and the IBEX 35 makes sense in order to analyse the effect of fragmentation on markets. The number of trades is a useful way to illustrate the significance of the impact of MiFID and high-frequency traders in European countries. FIGURE 7: AVERAGE DAILY NUMBER OF TRADES (BASED AT 1) 8 7 FTSE 100 CAC40 IBEX / / / / / / / / / / / / /03 Figure 7 indicates the relevant increase in high-frequency traders' activity, especially on the FTSE, which is much higher than in Spain, as would be expected from the difference in frictional costs. The real question is: how can one quantify the impact of fragmentation and high frequency trading on the markets?. To put things in context, Figure 1 indicates the significant increase in high-frequency traders' activity since the introduction of MiFID. One can question whether this change in market microstructure actually reduced volatility. Was this really the case? The following analysis focuses on this question. Volatility is one important point of market behaviour, as it is the random portion of the price formation process (PFP). It stems from business uncertainty on the listed firm, which becomes market uncertainty via the asset structure of the company. Thus, daily volatility represents the amount of risk taken by the investor. To attract investors, a stock with more volatility should offer a higher return. A lower bid-ask spread should imply lower intra-day volatility Microstructure implements a link between the market price and the fundamentals of a stock, via the price formation process. The more efficient the microstructure is, the lower the intra-day volatility of a stock. Intra-day volatility is the proxy for intra-day market risk and it is also part of the implicit costs of trading as a significant parameter in the market impact caused by a trade. Moreover, this link between intra-day volatility and trading costs is emphasised by the economically-founded relationship between volatility and the bid-ask spread. 15

16 We would therefore expect that a decreasing spread would imply decreasing volatility. However, while we have effectively noted a narrowing of bid-ask spreads after MiFID (mainly due to tick size changes), the decrease in volatility is less obvious (partly because of the financial crisis). Comparison is thus only significant between two markets that have faced similar market contexts. Our choice has been to focus on London and Spain. The former has clearly been impacted by fragmentation while the latter has more or less maintained its pre-mifid operating model. In order to compare the relative movements in intra-day volatility (which is computed using the 15-minute Garman-Klass estimator), we focus first on the difference between volatility for the FTSE 100 and that for the IBEX (Figure 8, upper chart): the difference remains positive (i.e. higher intra-day volatility for FTSE 100 stocks) and has been relatively stable since MiFID. One could argue that a decreasing slope could be assumed since August There is no evidence of a positive effect of fragmentation and high frequency trading on intra day volatility Although UK and Spanish stocks have both been exposed to the crisis, their respective exposure has not occurred at the same pace of with the same amplitude. To compensate for this effect, we used the differences between their intra-day volatility to daily volatility ratios (Figure 8, lower chart). The intra-day volatility is now applied relative to the overall uncertainty on each stock. As shown on the lower chart of Figure 8: the decrease in the gross difference since August 2008 has not been caused by fragmentation, but by the relatively slower recovery of Spain from the crisis (in terms of volatility). Once again, the difference has remained positive since MiFID: there is no evidence of a positive effect of fragmentation and high frequency trading on intra-day volatility. FIGURE 8: DIFFERENCE BETWEEN FTSE AND IBEX MOVING AVERAGE OVER 20 DAYS (BASE: 100) (UPPER CHART: INTRADAY VOLATILITY; LOWER CHART: INTRA-DAY VOLATILITY / DAILY VOLATILITY RATIO) /12/ /09/ /07/ /05/ /02/ /12/ /09/ /07/ /05/ /02/ /12/ /09/ /07/ /12/ /09/ /07/ /05/ /02/ /12/ /09/ /07/ /05/ /02/ /12/ /09/ /07/

17 II A new way to describe the liquidity discrepancy Entropy measures the level of "disorder" (i.e. fragmentation) in a system Entropy of the market microstructure It is obvious that if five trading venues are available for a given stock, the maximum level of fragmentation is attained when one-fifth of trades occur on each venue, and the lowest level of fragmentation when all deals are carried out on one venue only. As often in quantitative finance, it is worth looking at statistical thermodynamics (the source of the Brownian Motion and Black-Scholes formulae) to see if measurements are available to answer our needs. It is worth noting that mathematicians and physicists have a common methodology for measuring the dispersion of matter or the level of randomness of a phenomenon. This is called entropy. Entropy measures the level of "disorder" (i.e. fragmentation) in a system. If the consolidated market is considered a "system", it is perfectly ordered should all transactions take place on a single venue and its "disorder" increases with its fragmentation. For a given stock, when 70% of trades take place on one venue, 20% on a second and 10% on a third, its entropy will be 0.34, according to the following formula defined by Kolmogorov and Shannon: if the market share of the nth venue is Mn, then the entropy is: Entropy = M n log( M ) n n An entropy-based index: the CFI Entropy is often renormalized by a constant; we therefore propose here to use the maximum entropy possible for a market with N different venues, i.e. log(n). This gives us the following formula for normalized entropy: H 1 = M n n log( M n log( N) ) Normalized entropy is an overall measurement of the level of fragmentation of a stock or index Thus, for our example of a stock with a breakdown across three venues of 70/20/10, the renormalized entropy is 72%. More than a formula, this normalized entropy which we will refer to as CA Cheuvreux's Fragmentation Index (or CFI) can be used to produce meaningful analysis of the history of a stock s fragmentation, for instance compared to an index. 17

18 FIGURE 9: COMPARISON OF THE TRENDS IN CA CHEUVREUX'S FRAGMENTATION INDEX FOR THE CAC 40 AND THE CRÉDIT AGRICOLE S.A. SHARE CAC40 CAGR.PA /26/ /03/ /11/ /19/ /26/ /03/ /11/ /18/ /26/ /03/ /11/ /18/ /26/ /03/ /10/ /18/ /24/ /01/ /09/2008 Table 1 gives some reference values for CA Cheuvreux's fragmentation index (CFI). TABLE 1: REFERENCE VALUES FOR THE CFI Breakdown by venue (%) CFI value Breakdown by venue (%) CFI value 25/25/25/25 50/50 25/25/25/25/0 70/20/10 100% 70/20/5/5 63% 100% 64/24/5/5/1 60% 86% 70/20/10/0 58% 73% 80/10/5/5 51% It is worth noting that the CFI scores the fragmentation state of a liquidity pool relative to what it could be given the number of available venues. For instance, the CFI for a breakdown of 70/20/10 across three venues is 73%, while the same breakdown with four available venues (i.e. 70/20/10/0) yields 58%. The threshold to include a new MTF in the CFI has been set at 1% market share. Figure 10 below shows the past values of the CFI for the six main European indices monitored by the monthly CA Cheuvreux TAG liquidity indicators. The pan-european Euro STOXX 50 is, due to its composition, more fragmented than any country-related index: its CFI is over 65% split across 12 venues. On other indices, the impact of the end of the liquidity agreement of Turquoise in March 2008 can be read. Using such a metric is a good way to quantify the fragmentation on a segment in a single figure. Insofar as it resituates the essence of fragmentation in terms of its physical source of inspiration (entropy), the main fragmentation events will leave their mark on the CFI. 18

19 FIGURE 10: COMPARISON OF THE TRENDS IN CA CHEUVREUX'S FRAGMENTATION INDEX FOR MAIN EUROPEAN INDICES CAC40 FTSE 100 DAX BEL20 EURO STOXX 50 AEX Cheuvreux' fragmentation index (normalized entropy) /09 12/08 11/08 10/08 09/08 08/08 07/08 06/08 05/08 04/08 04/08 03/08 02/08 01/08 02/10 01/10 01/10 12/09 11/09 10/09 09/09 08/09 07/09 06/09 05/09 04/09 03/09 03/09 02/09 Because of its expressivity, the CFI will now be included in the monthly CA Cheuvreux TAG liquidity indicators. 19

20 III Dark pools, an alternative liquidity source Mechanism of dark liquidity pools As explained earlier in this report, a dark pool can be defined as an MTF that enables order matching without pre-trade transparency in respect of certain market rules. The main MTFs with significant market shares in the lit world (namely Chi-X, BATS, Turquoise and Nasdaq OMX) have launched their dark pools using the reference price waiver, which only permits orders to be crossed at the mid-point of the best bid and offer of a reliable source. The basic dark pool model is composed of two separate order books Small dark orders can only be crossed at mid-point, whereas large dark orders can participate in the price formation process To be consistent with MiFID rules and the principle that the smaller the quantity of an order, the more the owner has to be protected, a model with two distinct order books has been chosen: An integrated book The integrated book accepts all visible orders and hidden orders larger than the Large In Scale (LIS). An order is considered to be large in scale compared with the normal market size if it is equal to or larger than the minimum order size specified in Table 2 in Annex II of the MiFID Implementing Regulation. A dark or mid-point book This book accepts hidden orders smaller than the LIS, and matches only at the midpoint of the Primary Best Bid and Offer. The basic principle of order routing depends on the order size, "dark" LIS orders participate in the price formation process (i.e. they interact with lit orders and LIS orders in the integrated book), whereas "small" dark orders (under the LIS size threshold) can only be matched in the mid-point book. Adding to this standard, some dark pools follow specific rules in routing orders to each book, mainly depending on the order type. For instance, on Chi-X, LIS orders can also be routed to the dark (or mid-point) order book if the order is specified with a minimum quantity. To conclude on this overview, it is important to note that the four MTFs cited above only accept mid-peg dark orders, wherein limit price, minimum acceptable quantity, and IOC characteristics are allowed. 20

21 In-depth analysis of dark liquidity European dark pool market share Dark pools now account for 5% of non-otc liquidity In the overall European landscape, dark pools now represent approximately 5% of non-otc trades. The six biggest hidden liquidity pools are Chi-X, BATS, Turquoise, Liquidnet, Nomura and POSIT, and together they would represent more than 90% of dark traded volumes. To illustrate how fast this market share has grown, let s take a look at the proportion of notional that is executed in the dark for Chi-X and BATS over the past few months: FIGURE 11: PERCENTAGE OF DARK TRADES ON THE TWO MAIN DARK POOLS CHIX BATS /12/ /25/ /10/ /27/ /12/ /29/ /14/ /29/ /15/ /31/ /17/ /02/ /18/2009 Source: Chi-X and BATS It is worth noting that to achieve such market share growth, BATS typically uses aggressive pricing, and also offered a maker-taker fee schedule initially. Chi-X, BATS and Turquoise capture more than 60% of this liquidity To summarise, the dark pool landscape in Europe is currently evolving very fast and the three main MTFs CHI-X, BATS and Turquoise together represent more than 60% of European dark pool trades. Therefore, the data we will use to give an idea of the characteristics of this liquidity is drawn from the first three weeks of March 2010 on these three dark pools. Main characteristics of dark liquidity When inquiring into the possible benefits of trading on dark pools, two obvious reasons come to mind. The first is suggested by the very name of such destinations: being in the dark, i.e. hidden, should enable a reduction in "information leakage" (avoiding being front run) and market impact. The second reason is that dark pools enable investors to be slightly more aggressive but without paying the whole spread, hence trading at the midpoint. Therefore, the first question about the basic characteristics of dark pool liquidity would be whether investors are more likely to use dark pools in order to reduce market impact or to avoid paying at least half the spread. On the one hand, we know that market impact is driven by volume and volatility, so it would be interesting to look at the increase in dark pool volumes when volatility is high or when volumes are low. On the other hand, it should be possible to gauge the strength of the incentive of saving half the spread by observing 21

22 the joint distribution of dark volumes and spread. Indeed, if investors tend to use dark pools more when spreads are high, then this would mean that this liquidity is mainly driven by the appeal of earning part of the spread. Figures 12 to 15 show our initial attempt to look at such hypotheses. The methodology is simple: we gathered data (aggregated in one minutes slices) for the volume-weighted average spread (VWAS), volatility, and turnover traded in the dark. Then, for a single stock, we split the VWAS and the volatility into ten deciles and summed the turnover traded on dark pools according to the relevant deciles. Finally we summed these turnover on all stocks. Thus, we obtain a simple way to view the joint distribution of VWAS (or volatility) and dark pool turnover and the distribution of dark turnover conditional on VWAS or volatility. FIGURE 12: DISTRIBUTION OF MID-POINT TURNOVER ACCORDING TO THE DECILES OF VOLATILITY (CAC 40 STOCKS) FIGURE 13: DISTRIBUTION OF MID-POINT TURNOVER ACCORDING TO THE DECILES OF VOLATILITY (FTSE 100 STOCKS) 25% 25% 20% 20% Dark turnover distribution 15% 10% Dark turnover distribution 15% 10% 5 % 5 % 0 % Stock specific volatility decile 0 % Stock specific volatility decile FIGURE 14: DISTRIBUTION OF MID-POINT TURNOVER ACCORDING TO THE DECILES OF SPREAD (CAC 40 STOCKS) FIGURE 15: DISTRIBUTION OF MID-POINT TURNOVER ACCORDING TO THE DECILES OF SPREAD (FTSE 100 STOCKS) 15% 16% 14% 12% Dark turnover distribution 10% 5 % Dark turnover distribution 10% 8 % 6 % 4 % 2 % 0 % Stock specific VWAS decile 0 % Stock specific VWAS decile 22

23 Obviously, either the spread or the volatility is related to the volume traded in dark pools: the higher the volatility, the higher the volume traded in dark pools, with nearly one-quarter of overall turnover being traded one-tenth of the time when volatility is at its highest. But the conclusion is the same for the spread, with 15% of dark turnover being traded in the last decile of the spread. The conclusions that one might draw at this stage, with so few elements, would be that there seems to be a stronger relationship between volatility and dark turnover than between the spread and dark turnover. Nevertheless, it is well known that there is also a stronger relationship between volume and volatility than between spread and volume. Indeed, there is also a strong relationship between spread and volatility. The causality and the real dynamic of these relations are still unclear, but they are easy to illustrate. In Figures 16 to 18, we have data aggregated in fifteen minutes slices for Total SA (TOTF.PA) in February 2010 in order to chart some scatter plots illustrating these relationships. FIGURE 16: SPREAD AND VOLATILITY, FIGURE 17: VOLUME AND VOLATILITY, FIGURE 18: SPREAD AND VOLUME, TOTF.PA IN FEBRUARY 2010 TOTF.PA IN FEBRUARY 2010 TOTF.PA IN FEBRUARY 2010 Extreme events relationship log(σ) log(v) log(v) log(ψ) log(σ) log(ψ) Figures 16 and 17 show a strong relationship between spread and volatility on the one hand, and between volume and volatility on the other. But it is important to note that it does not imply such a clear link between spread and volume. Nevertheless, everyone knows that if an investor makes extensive use of market orders over a very short span of time, he will enlarge the spread. This can be seen in the extreme event circle. And indeed, there is a relationship between spread and volume, but mostly at these times. Nevertheless, the stronger link between volume and volatility than between spread and volatility implies that we must moderate the conclusions we would draw from Figures 12 to 15. Indeed, with only an extreme event relationship between spread and volume, we would now question why we see a relationship between dark turnover and spreads. A different approach to the question is to focus on the relationship between dark pools market share, and volatility or spread. This is especially interesting, because if this measure gives information about the relative incentive for the market to use dark pools more under certain market conditions, it would also give a simple way to split volume across lit and dark venues. In Figures 19 and 20, in order to illustrate the joint distribution of dark market share and volatility, we have proceeded as follows: we again used one minute data, and broke the volatility for a single stock into deciles. Then, for each stock of the index, we summed the dark turnover in each decile, as well as overall turnover, to obtain a dark market share for this stock in each volatility decile by dividing the former by the latter. One stock gives us one point in a decile of volatility. We repeated the computation for all the stocks of an index so that the following boxplots comprise as many points as stocks in the index. We 23

24 have added green points linked by a dark line to the boxplot, these are simply the average of the dark market share in the decile of volatility on the whole index. FIGURE 19: RELATION BETWEEN DARK POOL MARKET SHARE AND VOLATILITY ON CAC 40 STOCKS FIGURE 20: RELATION BETWEEN DARK POOL MARKET SHARE AND VOLATILITY ON FTSE 100 STOCKS 12% 5% 10% Dark pool market share 4% 3% 2% Dark pool market share 8 % 6 % 4 % 1% 2 % 0% 0 % Stock specific volatiliy deciles Stock specific volatiliy deciles Volatility is not yet the primary reason for investors to use dark pools The conclusion is clear: currently, volatility is not yet the strongest incentive for investors to use dark pools more. In Figures 21 and 22, we have followed exactly the same methodology, but instead of looking at volatility deciles we used the VWAS ones for the stocks. We also added the cross-stock median of the VWAS in basis points. This median is made across stocks and has no kind of weighting by the turnover of the stock, so these figures should be used with care as VWAS is very different for liquid and illiquid stocks; this is especially true when looking at the FTSE 100, which contains some very illiquid stocks. FIGURE 21: RELATION BETWEEN DARK POOL MARKET SHARE AND THE SPREAD ON CAC 40 STOCKS FIGURE 22: RELATION BETWEEN DARK POOL MARKET SHARE AND THE SPREAD ON FTSE 100 STOCKS 4 % 12% 3.5% 10% Dark pool market share 3 % 2.5% 2 % 1.5% 1 % 0.5% 0 % Dark pool market share 8 % 6 % 4 % 2 % 0 % Stock specific VWAS deciles Stock specific VWAS deciles 24

25 The higher the spread, the more liquidity in dark pools Contrary to volatility, the relationship between dark pool market share and spread is high. The median dark pool market share is indeed three times as high in the last decile of the spread than in the first decile. It is obvious from these figures that the higher the spread, the easier it will be to find some liquidity in dark pools. This is particularly true when looking at Figures 23 and 24 (below), which were built using the same methodology but using less liquid stocks from the SBF 120 (excluding CAC 40 stocks) and the FTSE 250. It is very easy to understand such behaviour by looking at the median spread of the last decile on the FTSE 250, which is as high as 77bp. Indeed, when an investor really needs to trade a stock that has a high spread and needs to trade fast, it is easily understandable that he will look to use dark pools before sending a market order. FIGURE 23: RELATION BETWEEN DARK POOL MARKET SHARE AND THE SPREAD ON SBF 120 STOCKS, EXCLUDING CAC 40 FIGURE 24: RELATION BETWEEN DARK POOL MARKET SHARE AND THE SPREAD ON FTSE 250 STOCKS 9% 8% 60% 50% Dark pool market share 7% 6% 5% 4% 3% 2% Dark pool market share 40% 30% 20% 10% 1% 0% Stock specific VWAS deciles 0 % Stock specific VWAS deciles Earning half the spread is better in a large spread context than in a small spread one There are many other good reasons for using dark pools than just using them before sending a market order. But we should agree that this is the most obvious one: earning half a spread when the latter is consequential. Nevertheless, it has many other advantages such as reducing the market impact. As the benefit of reducing market impact is more difficult to measure, it is not still clear that dark pools are being used in this way. For example, another incentive for using dark pools is when there are very few trades on lit pools, in which case dark pools are simply an additional liquidity pool; if the lit pools cannot provide enough liquidity, you have to try the dark ones. Such an effect can currently be observed only on the FTSE 100, which has always been the fastest to take advantage of new liquidity pools. This can be seen in Figure 25 (below), which is built using the same methodology as the previous charts, but the dark market share is conditional to overall turnover. This simply shows on the left of the figure that when the turnover on stocks is very low (first decile), there are relatively more trades being carried out in dark pools. 25

26 FIGURE 25: DARK MARKET SHARE CONDITIONAL ON OVERALL TURNOVER (FTSE 100 MARCH 2010, INTRA-STOCK ANALYSIS) FIGURE 26: DARK MARKET SHARE AND OVERALL TURNOVER (FTSE 100 AND FTSE 250 STOCKS MARCH 2010, CROSS-STOCK ANALYSIS) 16% 15 14% Dark pool market share 12% 10% 8 % 6 % 4 % Dark pools market share % 0 % Stock specific turnover deciles 0 liquidity indicator Dark pools are also an alternative means to find liquidity when it is very thin This could be a first sign of the advantage of using dark pools for reducing market impact (or at least the price impact), as there is no doubt that in such market conditions the market depth is very thin. This intra-stock analysis is confirmed by Figure 26 (above). On this figure, one point stands for one stock over the first three weeks of March We have represented the dark market share according to the liquidity of the stock. The stock liquidity indicator chosen here is the log of the turnover traded for this stock over the full period. There is obviously a variance effect, that is to say, when the turnover is low, then there is more variance in the dark market share. Nevertheless, we can also see that the lower a stock s liquidity, the more investors are likely to use dark pools in order to find liquidity. 26

27 IV An optimal liquidity capturing scheme CA Cheuvreux s CrossFire algorithm is an innovative liquidity-capturing scheme that allows investors to take advantage of all the new hidden liquidity available either in dark or lit pools. CrossFire s primary goal is to minimise information leakage by posting most of its size on dark pools. Its other purpose is to execute its size as quickly as possible while trying to execute at the best price available. This algorithm can be viewed as a "global limit order" on a short-time scale. CrossFire dynamically adapts its tactics depending on market conditions in order to minimise information leakage A glimpse at CrossFire s tactics CrossFire is self-adaptive in real time, and uses fill rates and market spread monitoring to dynamically rebalance the volume between trading venues and to adapt its distance from the reservation price. Based on specific trading rules (which are rooted in recent academic advances, an abstract of which is available as a preprint on the web: Optimal split of orders across liquidity pools: a stochastic algorithm approach by Gilles Pagès, Sophie Laruelle and Charles-Albert Lehalle), order placement on all trading destinations (MTFs, dark pools and the primary market) is optimised in order to: Avoid the limit price reaching the mid-point in order to be "in the money" most of the time in dark pools; Obtain the best price possible on MTFs and the primary market by taking advantage of a rapid widening in the spread; Use the more efficient pool and avoid gaming. The following figure (Figure 27) shows an example in which CrossFire proves very efficient. As the buy limit price is under the available aggressive quantity, CrossFire s passive placement on lit markets will take advantage of a rapid widening in the spread, and the mid-point placement maximises the possibility of being executed. This placement minimises information leakage (via the use of iceberg orders on the primary market and dark pools and) as well as rapid execution and price improvement (passive orders at best bid on MTFs). As a reminder, this figure is a static view of CrossFire, and depending on market conditions, it dynamically rebalances the volume between pools and places orders at the "best" distance from the limit price. 27

28 Limit price Mid price Bid price Ask price Order placement FIGURE 27: EXAMPLE OF A CROSSFIRE PLACEMENT TACTIC FOR A BUY ORDER Step 1: aggressive liquidity is out of the money - PM: Passive iceberg placement - MTF: Passive placement, more aggressive than on PM - Mid-point: Mid Peg order (with limit price) Primary market MTF Mid-point pool Step 2: market goes down - rebalancing of the passive placement (PM and MTF) - Mid Peg order is now in the money! Primary Market MTF Mid point pool CA Cheuvreux's CrossFire is efficient in minimising information leakage with 29% of the turnover matched on dark pools CrossFire efficiency The efficiency of minimising information leakage can be seen in the dark execution share. CA Cheuvreux s CrossFire executes 29% of its turnover in dark pools. Compared to the 5% dark market share across the main indices, CrossFire thus has impressive power to seek liquidity in dark pools. The same figures by index: 20% for the CAC 40 stocks, 23% for the AEX, 38% for the FTSE 100 and 45% for the SBF 120 (excluding CAC 40 stocks). This tends to bolster our previous results on agents state of mind. CrossFire manages to find more hidden liquidity on stocks with higher spreads and low lit liquidity, as investors tend to make more use of dark pools in order to capture part of the spread or as an alternative liquidity pool. CrossFire's efficiency can also be quantified by looking at the price improvement from the limit price (note that the limit price is always close to the best opposite at the launch of the order). 28

29 Figures 28 to 31 give the price improvement in basis points depending on the dark execution turnover share. Execution is divided into three classes: CA Cheuvreux's CrossFire, viewed as a "global limit order", is efficient on a short time scale No dark pool execution, so that the price improvement comes only from the execution of the passive (and sometimes aggressive) orders we posted on the various venues; Less than half of the order executed on dark pools. More than half of the order executed on dark pools. Whatever the marketplace, the higher the percentage of dark execution, the higher the price improvement. In fact, when none of the CrossFire execution is dark, price improvement is low. Conversely, when more than half the turnover is dark, the price improvement expands from 2bp to 10bp (the mean of price improvement for all stocks). To conclude, CrossFire can be a very useful tool for executing limit orders (close to best limits) on a short time scale. And this is true even for large orders because this algorithm can seek all liquidity available on any trading destination. FIGURE 28: PRICE IMPROVEMENT VERSUS DARK EXECUTION TURNOVER SHARE (ALL STOCKS) FIGURE 29: PRICE IMPROVEMENT VERSUS DARK EXECUTION TURNOVER SHARE (AEX STOCKS) % ]0 % ; 50%] >50% 0 0% ]0 % ; 50%] >50% FIGURE 30: PRICE IMPROVEMENT VERSUS DARK EXECUTION FIGURE 31: PRICE IMPROVEMENT VERSUS DARK EXECUTION TURNOVER SHARE (CAC 40) TURNOVER SHARE (FTSE 100) % ]0 % ; 50%] >50% 0 0% ]0 % ; 50%] >50% 29

30 Appendices Appendix 1: Glossary Price Formation Process The price formation process covers all events occurring during trading that result in a constantly evolving market price. Such events are, for instance: insertion of new orders or cancellation of orders, or matching of opposite orders. Walrasian Equilibrium An equilibrium between consumers and producers to set a perfect match between supply and demand. Trading Destinations 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). Primary Market Multilateral Trading Facility (MTF) Dark Pools Smart Order Router Also called new issue market. Securities are issued for the first time in this market. In this study, EURONEXT PARIS, the LONDON STOCK EXCHANGE and XETRA are examples of primary markets. A multilateral system operated by an investment firm or a market operator which brings together multiple third-party buying and selling interests in financial instruments in a way that results in a contract. Chi-X, Turquoise and BATS are the MTFs studied here. A trading destination not disclosing its order book. Trades are often reported with a delay. These venues are said to be adapted to trade large quantities without leaving a footprint in the market. A device routing an order across a given set of trading destinations according to a disclosed execution policy. It can split an order into smaller ones to spray all available destinations if needed. Execution Costs Execution costs are a mixture of fees, bid-ask spread, price impact, market impact, opportunity risk and market risk. Market Risk Price impact Market Impact Opportunity Risk Tick Size Measurement of the uncertainty in the evolution of the price. Impact of the volume of an aggressive order on the obtained average price. Possibly persistent impact of the volume of an aggressive order on the market price. Price of missing a transaction or obtaining a deteriorated price by not having placed an order on the adequate trading destination. The minimal difference allowed between two different prices. It is defined by the trading rules of each trading destination. VWAS: Volume Weighted Average Spread The mean of the bid-ask spread at each trade, weighted by the volume of the trade. Execution Benchmark Benchmarks define the target to be achieved during the trading of a large order. Implementation Shortfall PVOL: Percentage of Volume The average price of the order is compared to the market price before the beginning of trading. The volume curve of the order compared to that of the market for the same stock. VWAP: Volume Weighted Average Price The mean of traded prices weighted by traded volume. Cross trades are often excluded from this computation. 30

31 Appendix 2: Lists of stocks CAC 40 Stocks ACCOR ACCP.PA SAINT GOBAIN SGOB.PA AIR LIQUIDE AIRP.PA SCHNEIDER ELECTRIC SCHN.PA ALCATEL-LUCENT ALUA.PA VINCI SGEF.PA AXA AXAF.PA STMICROELECTRONICS STM.PA BNP PARIBAS ACT.A BNPP.PA SOCIETE GENERALE SOGN.PA BOUYGUES BOUY.PA TECHNIP TECF.PA CAP GEMINI CAPP.PA TOTAL TOTF.PA CARREFOUR CARR.PA UNIBAIL-RODAMCO UNBP.PA DANONE DANO.PA VALLOUREC VLLP.PA ESSILOR INTL. ESSI.PA DEXIA SA DEXI.BR FRANCE TELECOM FTE.PA SANOFI-AVENTIS SASY.PA L'OREAL OREP.PA ARCELORMITTAL ISPA.AS LAFARGE LAFP.PA EADS EAD.PA LAGARDERE S.C.A. LAGA.PA VEOLIA ENVIRON. VIE.PA LVMH LVMH.PA VIVENDI VIV.PA MICHELIN MICP.PA CREDIT AGRICOLE CAGR.PA PERNOD RICARD PERP.PA GDF SUEZ GSZ.PA PEUGEOT PEUP.PA ALSTOM ALSO.PA PPR PRTP.PA EDF EDF.PA RENAULT RENA.PA SUEZ ENVIRONNEMENT SEVI.PA SBF 120 excluding CAC 40 Stocks ALTRAN TECHN. ALTT.PA AREVA CI CEPFi.PA ATOS ORIGIN ATOS.PA TELEPERFORMANCE ROCH.PA BENETEAU CHBE.PA BONDUELLE BOND.PA FAURECIA EPED.PA EUROFINS SCIENT. EUFI.PA BIC BICP.PA STALLERGENES GENP.PA CARBONE LORRAINE CBLP.PA CNP ASSURANCES CNPP.PA CASINO GUICHARD CASP.PA BOURBON GPBN.PA SPERIAN PROTECTION SPEP.PA ALTEN LTEN.PA CIMENTS FRS ORD. CMFP.PA SOITEC SOIT.PA CLUB MEDITERRANEE CMIP.PA NEOPOST NPOS.PA DASSAULT SYSTEMES DAST.PA STERIA (GROUPE) TERI.PA EIFFAGE FOUG.PA IPSOS ISOS.PA ERAMET ERMT.PA SAFRAN SA SAF.PA EURAZEO EURA.PA NICOX NCOX.PA CGG VERITAS GEPH.PA TECHNICOLOR TCH.PA HAVAS EURC.PA EULER HERMES ELER.PA HERMES INTL HRMS.PA NEXANS NEXS.PA IMERYS IMTP.PA JC DECAUX SA. JCDX.PA INGENICO INGC.PA ORPEA ORP.PA KLEPIERRE LOIM.PA GECINA NOM. GFCP.PA WENDEL MWDP.PA ILIAD ILD.PA M6-METROPOLE TV MMTP.PA SES SESFd.PA NATIXIS CNAT.PA GEMALTO GTO.PA DERICHEBOURG DBG.PA BIOMERIEUX BIOX.PA PUBLICIS GROUPE SA PUBP.PA PAGESJAUNES PAJ.PA REMY COINTREAU RCOP.PA NEXITY NEXI.PA S.E.B. SEBF.PA SAFT S1A.PA SILIC SILP.PA MERCIALYS MERY.PA SODEXO EXHO.PA EUTELSAT COMMUNIC. ETL.PA TF1 TFFP.PA IPSEN IPN.PA THALES TCFP.PA LEGRAND LEGD.PA UBISOFT ENTERTAIN UBIP.PA ARKEMA AKE.PA VALEO VLOF.PA ADP ADP.PA VILMORIN & CIE VILM.PA EDF ENERGIES NOUV. EEN.PA ZODIAC AEROSPACE ZODC.PA SCOR SE SCOR.PA AIR FRANCE -KLM AIRF.PA REXEL RXL.PA ICADE ICAD.PA RHODIA RHA.PA MAUREL ET PROM MAUP.PA BUREAU VERITAS BVI.PA SECHILIENNE SIDEC SECH.PA GROUPE EUROTUNNEL GETP.PA FONC.DES REGIONS FDR.PA CFAO CFAO.PA 31

32 FTSE 100 Stocks DIAGEO PLC DGE.L STANDARD CHARTERED PLC STAN.L BARCLAYS PLC BARC.L HAMMERSON PLC HMSO.L LLOYDS BANKING GROUP PLC LLOY.L LIBERTY INTERNATIONAL PLC LII.L MARKS & SPENCER GROUP PLC MKS.L LONMIN PLC LMI.L UNILEVER PLC ULVR.L REXAM PLC REX.L BRITISH AIRWAYS PLC BAY.L SCHRODERS PLC SDR.L BP PLC BP.L SERCO GROUP PLC SRP.L PRUDENTIAL PLC PRU.L SEVERN TRENT PLC SVT.L RSA INSURANCE GROUP PLC RSA.L SHIRE PLC SHP.L VODAFONE GROUP PLC VOD.L SEGRO PLC SGRO.L KINGFISHER PLC KGF.L WM MORRISON SUPERMARKETS MRW.L BAE SYSTEMS PLC BAES.L INTERNATIONAL POWER PLC IPR.L WPP PLC WPP.L OLD MUTUAL PLC OML.L ASTRAZENECA PLC AZN.L ANGLO AMERICAN PLC AAL.L LEGAL & GENERAL GROUP PLC LGEN.L SCHRODERS PLC-NON VOTING SDRt.L PEARSON PLC PSON.L AMEC PLC AMEC.L TESCO PLC TSCO.L ANTOFAGASTA PLC ANTO.L SMITHS GROUP PLC SMIN.L CAIRN ENERGY PLC CNE.L HSBC HOLDINGS PLC HSBA.L TULLOW OIL PLC TLW.L LAND SECURITIES GROUP PLC LAND.L COBHAM PLC COB.L NEXT PLC NXT.L AGGREKO PLC AGGK.L ROYAL BANK OF SCOTLAND GROUP RBS.L ICAP PLC IAP.L BG GROUP PLC BG.L AUTONOMY CORP PLC AUTN.L NATIONAL GRID PLC NG.L CARNIVAL PLC CCL.L RIO TINTO PLC RIO.L GLAXOSMITHKLINE PLC GSK.L SAINSBURY (J) PLC SBRY.L COMPASS GROUP PLC CPG.L AVIVA PLC AV.L LONDON STOCK EXCHANGE GROUP LSE.L SMITH & NEPHEW PLC SN.L BT GROUP PLC BT.L BRITISH SKY BROADCASTING GRO BSY.L XSTRATA PLC XTA.L SCOTTISH & SOUTHERN ENERGY SSE.L INTERTEK GROUP PLC ITRK.L INVENSYS PLC ISYS.L BURBERRY GROUP PLC BRBY.L RECKITT BENCKISER GROUP PLC RB.L INVESTEC PLC INVP.L REED ELSEVIER PLC REL.L ALLIANCE TRUST PLC ATST.L BHP BILLITON PLC BLT.L RANDGOLD RESOURCES LTD RRS.L BRITISH AMERICAN TOBACCO PLC BATS.L INTERCONTINENTAL HOTELS GROU IHG.L ROLLS-ROYCE GROUP PLC RR.L VEDANTA RESOURCES PLC VED.L SAGE GROUP PLC/THE SGE.L G4S PLC GFS.L IMPERIAL TOBACCO GROUP PLC IMT.L ADMIRAL GROUP PLC ADML.L JOHNSON MATTHEY PLC JMAT.L ROYAL DUTCH SHELL PLC-A SHS RDSa.L ARM HOLDINGS PLC ARM.L INMARSAT PLC ISA.L 3I GROUP PLC III.L ROYAL DUTCH SHELL PLC-B SHS RDSb.L ASSOCIATED BRITISH FOODS PLC ABF.L PETROFAC LTD PFC.L BRITISH LAND CO PLC BLND.L KAZAKHMYS PLC KAZ.L BUNZL PLC BNZL.L STANDARD LIFE PLC SL.L CAPITA GROUP PLC CPI.L EXPERIAN PLC EXPN.L CENTRICA PLC CNA.L HOME RETAIL GROUP HOME.L MAN GROUP PLC EMG.L THOMAS COOK GROUP PLC TCG.L UNITED UTILITIES GROUP PLC UU.L TUI TRAVEL PLC TT.L WHITBREAD PLC WTB.L EURASIAN NATURAL RESOURCES ENRC.L WOLSELEY PLC WOS.L FRESNILLO PLC FRES.L SABMILLER PLC SAB.L CABLE & WIRELESS WORLDWIDE CWP.L 32

33 FTSE 250 Stocks RENTOKIL INITIAL PLC RTO.L AMLIN PLC AML.L COOKSON GROUP PLC CKSN.L FIDESSA GROUP PLC FDSA.L LADBROKES PLC LAD.L F&C ASSET MANAGEMENT PLC FCAM.L BBA AVIATION PLC BBA.L CONNAUGHT PLC CNT.L CABLE & WIRELESS COMMUNICATI CWC.L BIG YELLOW GROUP PLC BYG.L GKN PLC GKN.L UNITE GROUP PLC UTG.L COLT TELECOM GROUP SA COLT.L DOMINO'S PIZZA UK & IRL PLC DOM.L UNITED BUSINESS MEDIA LTD UBM.L SDL PLC SDL.L DSG INTERNATIONAL PLC DSGI.L MELROSE RESOURCES PLC MRS.L LOGICA PLC LOG.L BALFOUR BEATTY PLC BALF.L PREMIER FARNELL PLC PFL.L SPIRENT COMMUNICATIONS PLC SPT.L MISYS PLC MSY.L GENUS PLC GENS.L HAYS PLC HAYS.L DIMENSION DATA HOLDINGS PLC DDT.L AEGIS GROUP PLC AEGS.L EASYJET PLC EZJ.L CLOSE BROTHERS GROUP PLC CBRO.L MICHAEL PAGE INTERNATIONAL MPI.L DAILY MAIL&GENERAL TST-A NV DMGOa.L JPMORGAN JAPANESE INV. TRUST JFJ.L FIRSTGROUP PLC FGP.L SYNERGY HEALTH PLC SYR.L TOMKINS PLC TOMK.L RIT CAPITAL PARTNERS PLC RCP.L TRINITY MIRROR PLC TNI.L POLAR CAPITAL TECHNOLOGY TR PCT.L SSL INTERNATIONAL PLC SSL.L FOREIGN & COLONIAL INVEST TR FRCL.L STAGECOACH GROUP PLC SGC.L PETROPAVLOVSK PLC POG.L TATE & LYLE PLC TATE.L HMV GROUP PLC HMV.L IMI PLC IMI.L FIDELITY EUROPEAN VALUES PLC FEV.L PROVIDENT FINANCIAL PLC PFG.L PUNCH TAVERNS PLC PUB.L RANK GROUP PLC RNK.L WOOD GROUP (JOHN) PLC WG.L MILLENNIUM & COPTHORNE HOTEL MLC.L WILLIAM HILL PLC WMH.L PENNON GROUP PLC PNN.L BLACKROCK WORLD MINING TRUST BRWM.L ARRIVA PLC ARI.L FIDELITY SPECIAL VALUES PLC FSV.L DE LA RUE PLC DLAR.L SCOTTISH MORTGAGE INV TR PLC SMT.L GAME GROUP PLC GMG.L EDINBURGH INVESTMENT TRUST EDIN.L ELECTROCOMPONENTS PLC ECM.L MERCHANTS TRUST PLC MRCH.L NATIONAL EXPRESS GROUP PLC NEX.L ABERFORTH SMALLER COS-ORD ASL.L NORTHERN FOODS PLC NFDS.L BANKERS INVESTMENT TRUST BNKR.L PACE PLC PIC.L BRITISH ASSETS TRUST PLC-ORD BSET.L MOTHERCARE PLC MTC.L BRITISH EMP SEC AND GEN-ORD BTEM.L CHARTER INTERNATIONAL PLC CHTR.L DUNEDIN INCOME GROWTH INV TR DIG.L DOMINO PRINTING SCIENCES PLC DOPR.L ELECTRA PRIVATE EQUITY PLC ELTA.L ABERDEEN ASSET MGMT PLC ADN.L MERCANTILE INVESTMENT TRUST MRCM.L ASHTEAD GROUP PLC AHT.L JPMORGAN AMERICAN INVEST TST JAM.L HUNTING PLC HTG.L MONKS INVESTMENT TRUST PLC MNKS.L BARR (A.G.) PLC BAG.L MURRAY INCOME TRUST PLC MUT.L BARRATT DEVELOPMENTS PLC BDEV.L MURRAY INTERNATIONAL TR-O MYI.L BERKELEY GROUP HOLDINGS BKGH.L TR PROPERTY INVESTMENT TRUST TRY.L BODYCOTE PLC BOY.L TEMPLE BAR INVESTMENT TRUST TMPL.L BSS GROUP PLC BTSM.L SCOTTISH INVESTMENT TRUST SCIN.L BROWN (N) GROUP PLC BWNG.L TEMPLETON EMERGING MARKETS-O TEM.L BELLWAY PLC BWY.L SVG CAPITAL PLC SVI.L CHEMRING GROUP PLC CHG.L BEAZLEY PLC BEZG.L CHLORIDE GROUP PLC CHLD.L EDINBURGH DRAGON TRUST PLC EDDR.L CALEDONIA INVESTMENTS PLC CLDN.L LAW DEBENTURE CORP PLC-ORD LWDB.L RATHBONE BROTHERS PLC RAT.L GENESIS EMERGING MARKETS GENE.L CRODA INTERNATIONAL PLC CRDA.L JPMORGAN ASIAN INVESTMENT TR JAI.L RESTAURANT GROUP PLC RTN.L IMPAX ENVIRONMENTAL MARKETS IMPX.L CRANSWICK PLC CWK.L JPMORGAN EURO FLEDGELING-ORD JFF.L DAIRY CREST GROUP PLC DCG.L JPMORGAN EMERGING MKTS TRST JMG.L DAEJAN HOLDINGS PLC DJAN.L PERPETUAL INCOME & GROWTH-O PLI.L DAVIS SERVICE GROUP PLC DVSG.L CITY OF LONDON INVESTMENT TR CTY.L DERWENT LONDON PLC DLN.L JPMORGAN INDIAN INV TRUST JII.L FENNER PLC FENR.L MITCHELLS & BUTLERS PLC MAB.L GREENE KING PLC GNK.L NORTHUMBRIAN WATER GROUP PLC NWG.L GREAT PORTLAND ESTATES PLC GPOR.L KESA ELECTRICALS PLC KESA.L GREGGS PLC GRG.L YELL GROUP PLC YELL.L GRAINGER PLC GRI.L MELROSE PLC NYN.L HELICAL BAR PLC HLCL.L REGUS PLC RGU.L HALMA PLC HLMA.L HENDERSON GROUP PLC HGGH.L INCHCAPE PLC INCH.L ITV PLC ITV.L LAIRD PLC LRD.L CSR PLC CSR.L MORGAN CRUCIBLE COMPANY PLC MGCR.L CATLIN GROUP LTD CGL.L MEGGITT PLC MGGT.L DIGNITY PLC DTY.L PZ CUSSONS PLC PZC.L HALFORDS GROUP PLC HFD.L PREMIER OIL PLC PMO.L PREMIER FOODS PLC PFD.L PERSIMMON PLC PSN.L BOOKER GROUP PLC BOK.L ROTORK PLC ROR.L STOBART GROUP LTD STOB.L RPS GROUP PLC RPS.L AFREN PLC AFRE.L RENISHAW PLC RSW.L IG GROUP HOLDINGS PLC IGG.L SHAFTESBURY PLC SHB.L MICRO FOCUS INTERNATIONAL MCRO.L SOCO INTERNATIONAL PLC SIA.L FILTRONA PLC FLTR.L 33

34 FTSE 250 Stocks DS SMITH PLC SMDS.L WITAN INVESTMENT TRUST PLC WTAN.L ST. MODWEN PROPERTIES PLC SMP.L PARTYGAMING PLC PRTY.L SIG PLC SHI.L ECOFIN WATER & POWER OPP-ORD ECW.L SPIRAX-SARCO ENGINEERING PLC SPX.L 888 HOLDINGS PLC 888.L TAYLOR WIMPEY PLC TW.L HIKMA PHARMACEUTICALS PLC HIK.L VT GROUP PLC VTG.L STHREE PLC STHR.L WEIR GROUP PLC/THE WEIR.L HANSTEEN HOLDINGS PLC HSTN.L MARSTON'S PLC MARS.L BRITVIC PLC BVIC.L SAVILLS PLC SVS.L LANCASHIRE HOLDINGS LTD LRE.L TRAVIS PERKINS PLC TPK.L DEXION ABSOLUTE LTD-SHS NPV DAB.L SPECTRIS PLC SXS.L DRAX GROUP PLC DRX.L MITIE GROUP PLC MTO.L QINETIQ GROUP PLC QQ.L CARPETRIGHT PLC CATVU.L RIGHTMOVE PLC RMV.L EUROMONEY INSTL INVESTOR PLC ERM.L HSBC INFRASTRUCTURE CO HICL.L BABCOCK INTL GROUP PLC BAB.L DEBENHAMS PLC DEB.L DANA PETROLEUM PLC DNX.L WH SMITH PLC SMWH.L ST JAMES'S PLACE PLC SJP.L UK COMMERCIAL PROPERTY TRUST UKCM.L GALIFORM PLC GFRM.L LAMPRELL PLC LAM.L ATKINS (WS) PLC ATKW.L ASHMORE GROUP PLC. ASHM.L IMAGINATION TECH GROUP PLC IMG.L DUNELM GROUP PLC DNLM.L FORTH PORTS PLC FPT.L HOCHSCHILD MINING PLC HOCM.L SENIOR PLC SNR.L INTERNATIONAL PUBLIC PARTNER INPP.L WETHERSPOON (J.D.) PLC JDW.L BLUEBAY ASSET MANAGEMENT BBAY.L VICTREX PLC VCTX.L SALAMANDER ENERGY PLC SMDR.L BTG PLC BGC.L TULLETT PREBON PLC TLPR.L AQUARIUS PLATINUM LTD AQP.L GEM DIAMONDS LTD GEMD.L ITE GROUP PLC ITE.L SPORTS DIRECT INTERNATIONAL SPD.L ROBERT WISEMAN DAIRIES PLC RWD.L 3I INFRASTRUCTURE PLC 3IN.L CARILLION PLC CLLN.L BH MACRO LTD BHMG.L KELLER GROUP PLC KLR.L BH MACRO LTD - EURO BHMGx.L GO-AHEAD GROUP PLC GOG.L XCHANGING PLC XCH.L REDROW PLC RDW.L WELLSTREAM HOLDINGS PLC WSML.L JKX OIL & GAS PLC JKX.L HARGREAVES LANSDOWN PLC HRGV.L BREWIN DOLPHIN HOLDINGS PLC BRW.L BLUECREST ALLBLUE FUND LD ORD NPV GBP BABS.L INTERMEDIATE CAPITAL GROUP ICP.L TALVIVAARA MINING CO PLC TALV.L SHANKS GROUP PLC SKS.L EAGA PLC EAGA.L INFORMA PLC INF.L FERREXPO PLC FXPO.L HISCOX LTD HSX.L MONDI PLC MNDI.L MCBRIDE PLC MCB.L INTERNATIONAL PERSONAL FINAN IPF.L ENTERPRISE INNS PLC ETI.L TR PRPTY INVESTMENT-SIGMA TRYx.L BRIT INSURANCE HOLDINGS NV BRE.L MONEYSUPERMARKET.COM MONY.L BOVIS HOMES GROUP PLC BVS.L TELECITY GROUP PLC TCY.L COMPUTACENTER PLC CCC.L HANSEN TRANSMISSIONS INT HSNT.L HOMESERVE PLC HSV.L HERITAGE OIL PLC HOIL.L ULTRA ELECTRONICS HLDGS PLC ULE.L BH GLOBAL LTD-GBP SHRS BHGG.L AVEVA GROUP PLC AVV.L BH GLOBAL LTD-USD SHRS BHGGu.L KIER GROUP PLC KIE.L RESOLUTION LTD RSL.L JARDINE LLOYD THOMPSON GROUP JLT.L F&C COMMERCIAL PROPERTY TRUS FCPTL.L PARAGON GROUP COMPANIES PLC PARA.L GARTMORE GROUP LTD GRTR.L TALKTALK TELECOM GROUP TALK.L IBEX 35 Stocks ACCIONA ANA.MC OBRASCON HUARTE LAIN OHL.MC ACS ACTIVIDADES CONST.Y SERVICIOS ACS.MC RED ELECTRICA CORPORACION SA REE.MC BANCO ESPANOL DE CREDITO BTO.MC BANCO BILBAO VIZCAYA ARGENTARIA BBVA.MC BANKINTER BKT.MC GAMESA CORPORACION TECNOLOGICA GAM.MC BANCO POPULAR ESPANOL POP.MC IBERIA IBLA.MC ENDESA ELE.MC IND. DE DISENO TEXTIL INDITEX ITX.MC FOMENTO CONSTR.Y CONTRATAS(FCC) FCC.MC BANCO SANTANDER SA SAN.MC GAS NATURAL SDG GAS.MC ENAGAS ENAG.MC IBERDROLA IBE.MC ABERTIS INFRAESTRUCTURAS SERIE A ABE.MC MAPFRE S.A. MAP.MC GESTEVISION TELECINCO TL5.MC REPSOL YPF REP.MC GRIFOLS GRLS.MC TELEFONICA TEF.MC FERROVIAL SA FER1.MC SACYR VALLEHERMOSO SVO.MC TECNICAS REUNIDAS TRE.MC ACERINOX ACX.MC BOLSAS Y MERCADOS ESPANOLES BME.MC EBRO PULEVA EVA.MC BANCO DE SABADELL NEW SABE.MC ABENGOA ABG.MC CRITERIA CAIXACORP SA CRIT.MC INDRA SERIE A IDR.MC ARCELOR MITTAL MTSbl.MC IBE.RENOVABL IBR.MC 34

35 Other publications EXECUTION SERVICES A bi-annual publication that analyses the issues surrounding liquidity Access, fragmentation and optimal trading destinations. Your comprehensive guide to our Execution Services offer; Sales Trading, Direct Market Access, Algorithmic Trading, Global Portfolio Trading, Contracts for Difference An in-depth look at our 16 customisable algorithmic strategies, to meet any investment parameters. Outlines the key benefits and logistics of signing a Commission Sharing Agreement wit CA Cheuvreux. Note: This document is also available in French. A quarterly newsletter investigating the latest issues in the Alternative Execution arena with articles from our industry experts. A guide to help you navigate the liquidity maze by detailing the trading conditions of the execution venues we trade on. 35