Order-Routing Binck-Alex / TOM. Flash Boys in The Netherlands



Similar documents
SAXO BANK S BEST EXECUTION POLICY

High-frequency trading: towards capital market efficiency, or a step too far?

BEST EXECUTION AND ORDER HANDLING POLICY

Options on. Dow Jones Industrial Average SM. the. DJX and DIA. Act on the Market You Know Best.

Hargreaves Lansdown Spread Betting/ Hargreaves Lansdown CFDs Trade & Order Execution Policy

OTC Options as Qualified Covered Call Options

FEATURES LIST OF THE SYSTEMS FOR SMART ORDER ROUTING AND THE APPLICABLE TERMS AND CONDITIONS

How To Trade Against A Retail Order On The Stock Market

The Need for Speed: It s Important, Even for VWAP Strategies

EXANE GROUP EXECUTION POLICY

Algorithmic and advanced orders in SaxoTrader

EUROPEAN SECURITIES AND MARKETS AUTHORITY (ESMA) CONSULTATION PAPER

MiFID II Trading Venues And High Frequency Trading

A case analysis of critiques on high frequency trading

Plus500CY Ltd. Order Execution Policy

3 Balance Sheets of Banks and Insurance Companies

ELECTRONIC TRADING GLOSSARY

BEST EXECUTION POLICY - FINANCIAL INSTRUMENTS

Options Pre-Trade and Post-Trade Risk Controls. NYSE Amex Options NYSE Arca Options. nyse.com/options

BEAR: A person who believes that the price of a particular security or the market as a whole will go lower.

General Forex Glossary

Goldman Sachs s Standard Spot Foreign Exchange Terms of Dealing

ORDER EXECUTION POLICY

Trade and Order Execution Policy for Retail and Professional Clients

Buy-ins, how they work, and the challenge of CSDR An ICMA briefing note July 2015 (updated) Introduction. How this paper is organized

Derivative Users Traders of derivatives can be categorized as hedgers, speculators, or arbitrageurs.

BEST EXECUTION AND ORDER HANDLING POLICY

Transaction Reporting - What is Changing Under MiFID II

Liquidity in U.S. Treasury spot and futures markets

White Paper Electronic Trading- Algorithmic & High Frequency Trading. PENINSULA STRATEGY, Namir Hamid

Investors Exchange (IEX) Building A Market that Works for Investors. September 9, 2015

XIV. Additional risk information on forward transactions in CFDs

Transaction Cost Analysis and Best Execution

G100 VIEWS HIGH FREQUENCY TRADING. Group of 100

General Risk Disclosure

FUNDING INVESTMENTS FINANCE 738, Spring 2008, Prof. Musto Class 4 Market Making

Pictet Asset Management: Summary of Best Execution Policy

OPTION TRADING 101. Session Objectives: Disclaimers: Expand knowledge of available tools Overview of Option Trading

PUREDMA TRADING MANUAL

FIN FINANCIAL INSTRUMENTS SPRING Options

Dated January 2015 Advanced Execution Services. Crossfinder User Guidelines Asia Pacific

2012 Fidessa group plc

High Frequency Trading Turns to High Frequency Technology to Reduce Latency

Commodity Options as Price Insurance for Cattlemen

PACIFIC PRIVATE BANK LIMITED S BEST EXECUTION POLICY (ONLINE TRADING)

FINAL NOTICE. Michael Coscia. Date of Birth: 7 July Date: 3 July 2013 ACTION

Centralised Portfolio Management

ATF Trading Platform Manual (Demo)

Increased Scrutiny of High-Frequency Trading

Chapter 12 Practice Problems

Setting the Scene. FIX the Enabler & Electronic Trading

Contents. 01 An Introduction to DMA trading within What is DMA? Benefits of DMA

3.1 Saxo Capital Markets identifies and seeks to obtain the most favorable terms reasonably available when executing an order on behalf of a client.

Robert Bartlett UC Berkeley School of Law. Justin McCrary UC Berkeley School of Law. for internal use only

How to Win the Stock Market Game

RISK WARNING. Retail transactions conducted with NatureForex are not insured by any deposit insurance of any kind.

Understanding Stock Options

BEST INTEREST AND ORDER EXECUTION POLICY

Understanding ETF liquidity and trading

POLICY STATEMENT Q-22

ABN AMRO Order Execution Policy

Market Microstructure: An Interactive Exercise

A GUIDE TO THE STRUCTURE, MARKET TERMINOLOGY AND ORDER EXECUTION OF THE LONDON METAL EXCHANGE

Financial Markets. Itay Goldstein. Wharton School, University of Pennsylvania

Algorithmic trading Equilibrium, efficiency & stability

Understanding Margins

BUSINESS TERMS FOR SECURITIES TRADING via DNB Trade

In effect CFD s are financial derivatives, originally known as Traded Options, that allow traders to take advantage of prices moving up (long

THE STOCK MARKET GAME GLOSSARY

SAVI TRADING 2 KEY TERMS AND TYPES OF ORDERS. SaviTrading LLP 2013

High-frequency traders have

There are two types of options - calls and puts.

AUTOMATED TRADING RULES

Understanding Margins. Frequently asked questions on margins as applicable for transactions on Cash and Derivatives segments of NSE and BSE

October 2003 UNDERSTANDING STOCK OPTIONS

How To Know Market Risk

Chapter 3. How Securities are Traded

Overview of the LOYAL3 Platform: How It Works + Specific Risks

Schneider. Foreign Exchange. BBI Finance is an introductory broker of Schneider Foreign Exchange Ltd.

Powerful tools for investing, speculating or hedging

Anti-Gaming in the OnePipe Optimal Liquidity Network

Bodie, Kane, Marcus, Perrakis and Ryan, Chapter 3

ORDER EXECUTION AND CONFLICTS OF INTEREST MANAGEMENT POLICIES

FTS Real Time Client: Equity Portfolio Rebalancer

Introduction to Equity Derivatives on Nasdaq Dubai NOT TO BE DISTRIUTED TO THIRD PARTIES WITHOUT NASDAQ DUBAI S WRITTEN CONSENT

RISK DISCLOSURE NOTICE

Strategies in Options Trading By: Sarah Karfunkel

Transcription:

Order-Routing Binck-Alex / TOM Flash Boys in The Netherlands

Executive Summary... 3 Introduction... 4 Research Question... 5 Order Routing... 6 Performed Tests... 7 Test Scenarios... 8 Results and Interpretation: The Flash Boys Scenario... 10 Flash Boys... 10 Binck/TOM comparison to Flash Boys... 11 Conclusion... 12 Appendix 1 Appendix 2 - Page 2 -

Executive Summary In this report, we investigate the assertion that the best execution services provided by Binck/TOM are better for end users than the method adopted by DeGiro the latter of which involves sending orders directly to Euronext, and the former to multiple exchanges. Trading Euronext shares listed on several exchanges was made possible in Europe by MiFID regulations which allow competition between exchanges. To this end, we construct a series of tests wherein we send sufficiently large orders to the market, so that they are routed to multiple trading platforms, thereby taking advantage of the above-mentioned best execution services at Binck/TOM. This procedure involves sending limit orders with the limit price set slightly worse than the current market price, with the aim of immediate execution potentially into the book depth (i.e. deeper than the best bid-offer BBO ). We then carefully analyze the performance of these executions, and uncover the existence of persistent flash trading, wherein other parties orders are executed almost immediately before our own at more favourable prices. Details regarding each of the tests that we have undertaken are documented (in the presence of journalists from Nieuwsuur and the Financieele Dagblad ) and set out in this report. Our tests uncover strong evidence of poor execution quality at unfavourable prices in almost all of the cases we have investigated (i.e. paying more for the purchase, and/or receiving less for the sale of, a stock). The fact that Binck s Smart Order Router delivers sub-optimal execution performance for customers, in addition to the evidence we present of pervasive flash trading, bears a strong analogy to Michael Lewis book Flash Boys a best-seller that has led the US Congress, and increasingly Europe, to scrutinize more deeply the controversial practice of flash trading. We herein compare the practices espoused by Binck to that of Flash Boys, and suggest that the situation they have knowingly created through TOM, which puts customers at a disadvantage, reflects many aspects of that book. For the benefit of our clients, DeGiro will certainly not be making use of the (current version of) TOM s best-execution services. There is a trend across European markets wherein brokers and market makers increasingly cooperate. DeGiro s position in this environment is one of extreme caution towards utilizing these services, and in particular an aversion to facilities which involve Smart Order Routing. - Page 3 -

Introduction New MiFID legislation in 2007 made competition between trading venues possible in Europe. Shares that had previously traded almost exclusively on a single stock exchange could now be traded on multiple markets. A broker such as DeGiro must, when executing orders on behalf of customers, choose how they execute these orders across the various exchanges and trading platforms. Following its partnership with Euronext, DeGiro has been criticized by Binck for the fact that orders on Euronext-listed stocks are routed by DeGiro to Euronext, and do not utilize the best execution services offered by TOM. This report sets out to determine whether it is indeed in the best interests of customers to route orders to TOM s "best execution services" facility. The main issue under investigation thus is whether the services offered by Binck specifically, routing orders on Euronext products to TOM, Chi-X, and Euronext offer any advantages (or disadvantages) to their end users as compared to DeGiro s approach. The first part of this report describes the issues under investigation and the central research theme in more detail. Subsequently, we describe the series of tests we carry out (buy and sell orders in 5 stocks) which are aimed at determining the quality of execution services offered by Binck. Thereafter, we present the results of these tests, along with careful analysis of the market conditions and order book situation at the time of each trade/order. This is facilitated by Bloomberg, via which we pay close attention to the sequence of ticks across all relevant trading venues, to determine precisely what happens in each market, and in what order, from the placement of our orders to their execution. Detailed results of all of these tests can be found in this report; a summary of the details of each test can be found in Appendix 1; while Appendix 2 explores our first test scenario in greater detail, with timestamp resolutions at the microsecond level. Our analysis shows conclusive evidence that, when routed via the best execution services offered by TOM, we encounter sub-optimal execution quality, and more worryingly, evidence of malpractice in the form of flash trades, wherein other parties orders are executed almost immediately before our own at more favourable prices. We draw a direct parallel between this situation and Michael Lewis book Flash Boys, and suggest that the experiences of Binck s / TOM s customers resembles the situation described in the book. We conclude definitively that Binck s customers are not better off with the use of TOM "best execution services". - Page 4 -

Research Question In this report, we investigate the effectiveness and quality of the order routing facility provided by Binck / TOM with regards to the execution of orders on multiple trading venues. Specifically, Binck s facility allows the routing of orders on Euronext-listed products via TOM, Chi-X, Euronext. It is repeatedly asserted in Binck-related media publications that this facility benefits Binck s customers by improving the execution quality of their orders. The aim of this study is to determine whether in fact this routing facility improves the execution quality of Binck / TOM s customer orders. The stocks utilized in this study are all Euronext-listed. This is particularly relevant, and Euronext-listed securities constitute a substantial part of order flow for both DeGiro and Binck. - Page 5 -

Order Routing Online brokers receive orders from customers and ensure that these are executed at an exchange. Since the MiFID legislation, it has become possible to trade the same stock on multiple venues, hence rival platforms compete for order flow. For example, a company such as Philips, whose shares have traditionally only traded on Euronext, may now be traded via several platforms. Under this legislation, it is also obligatory for brokerages to provide the fairest and best possible order-routing facility for the benefit of their customers. The brokerage itself has a (possibly opposite) interest in this matter. Specifically, it would want to the extent possible, route customer orders to venues that would charge the brokerage the lowest fees, so as to maximize its profit margin on carrying that order. If a brokerage offers the facility of connecting to multiple trading venues for a given stock, it would have to have a system through which customer orders can be optimally routed, and an established order-routing methodology. This system is known as Smart Order Router. Within the context of this study, we look at the Smart Order Router provided by Binck / TOM to route orders on Euronext-listed products across the multiple venues on which they may be traded. This is illustrated in the figure below: Figuur 1. Order-routing flowchart for Binck / Tom 1. The "best-execution service" stated in Figure 1 performs the function of the Smart Order Router. This facility must ensure the most efficient and beneficial execution for the customer s order. Stated on BinckBank s website is as follows: "TOM compares different exchanges and trading platforms within a fraction of a second and places your order where it can be executed at the best price." 1 Source https://www.binck.nl/tom. Figure is somewhat outdated, as it now also includes Chi-X. - Page 6 -

Performed Tests The tests we carry out consist of both buy and sell orders in a number of Euronext-listed stocks. The stocks chosen for this exercise are as follows: Solvay (4 test orders) Unibail Rodamco (2 test orders) Heineken (2 test orders) Randstad (2 test orders) L Oreal (2 test orders) The above stocks span a wide range of daily liquidity. In particular, both Solvay and Unibail are relatively illiquid stocks. The inclusion of these into our test environment provides an excellent opportunity to clearly see the sequence of market events around the execution of our orders. Furthermore, the inclusion of Heineken and Randstad allow us to apply our Research Question to liquid stocks as well, hence ensuring the liquidityrepresentativeness of our sample. In order to determine the size and limit price of our orders, we first examine the state of the order book at the individual trading venues. By way of example, suppose that the market is as follows: Exchange Bid volume Bid Price Ask Price Ask Volume Euronext 100 10.10 10.15 500 150 10.05 10.20 500 10.25 1000 Chi-X 100 10.10 10.15 200 150 10.05 10.20 100 10.25 300 Tom 200 10.10 10.15 200 Suppose for this example that we wish to purchase a quantity of 1000 shares. Across the exchanges, there are 900 shares available at a price of 10.15 (Chi-X 200, TOM 200, and Euronext 500). The remaining 100 shares may be purchased, but they must be done so at a higher price (10.20 in this case). Suppose that our test order consists of buying 1000 shares at a limit price of 10.30. If this order was sent directly to Euronext, it would be executed at an average price of 10.175 (500 @ 10.15 and 500 @ 10.20). The question now is whether making use of a Smart Order Router, which is capable of routing the order to multiple venues, provides an execution at a lower price. This is the central research theme of this report. Each of the tests involve sending orders larger than the combined quantity available across all trading venues, and then assessing the optimality of the executions we receive. Specifically, an optimal execution would achieve the best price across all exchanges, thus effectively giving us efficient access to the aggregate order book this is the logical interpretation of the claim made by Binck. An optimal execution clearly implies that the customer is better off utilizing the Smart Order Router facility; and if not, then the customer is better off having their order routed directly to a single (main exchange). In this report, we calculate the value gained/lost to customers utilizing either approach. - Page 7 -

Test Scenarios We have sent the following orders to Binck/TOM: Time Order Details Observations Buy 2000 shares Solvay @ 113.30 1 18 nov 2014 13:26:38 Annex, p. 1 2 18 nov 2014 13:48:28 Annex, p. 11 3 25 nov 2014 10:39:19 Annex, p. 21 4 25 nov 2014 11:06:12 Annex, p. 33 5 25 nov 2014 12:11:07 Annex p. 42 6 25 nov 2014 12:26:10 Annex p. 52 7 25 nov 2014 11:44:42 Annex p. 64 8 25 nov 2014 12:05:00 Annex p. 73 9 25 nov 2014 12:40:04 Annex p. 82 10 25 nov 2014 12:51:41 Annex p. 90 11 25 nov 2014 11:14:29 Annex p. 99 12 25 nov 2014 11:14:29 Annex p. 106 Sell 2000 shares Solvay @ 113.20 Buy 2000 shares Solvay @ 111.90 Sell 2000 shares Solvay @ 112.00 Buy 3000 shares L Oréal @136.00 Sell 3000 shares L Oréal @ 135.80 Buy 5000 shares Randstad @ 39.95 Sell 5000 shares @ 39.85 Buy 4000 shares Heineken @ 62.25 Sell 5000 shares Heineken @ 61.90 Buy 1000 shares Unibail @ 208.50 Sell 1000 shares Unibail @ 207.80 1. On Euronext, trades executed against our order are immediately preceded by others at better prices (flash trade) 2. Sequential executions against our order are clearly interrupted by other executions (not ours). 3. This test is presented in great detail, since we have received granular data from TOM and Euronext for this one. 1. Executions against other orders immediately before ours on Chi-X (flash trade) 2. Executions against other orders immediately before ours on Euronext (flash trade) 1. The volume available at 111.80 on Chi-X and the next level was bought before our order (flash trade). 2. The volume available on Euronext at 111.80 was executed against another party s order (flash trade). We did not buy all shares available at 111.85. No details in this transaction. Striking that we only see feedback on this transaction from TOM. 1. We buy 147 shares on Chi-X, but worse than the market. Another party s order is executed at the then-current price (flash trade). 2. We buy the last shares on Chi-X at 136.- while there exists a better price on Euronext (sub-optimal order routing) 3. On Euronext, we buy only 89 shares at 135.95 while more volume is available no reason to buy at Chi-X. (suboptimal order routing) 1. On Chi-X, another party s order is executed before ours. (flash trade) 2. On Chi-X we sell 469 shares at 135,90, while we could have sold on Euronext at 135,95 (sub-optimal order routing) 1. On another order was executed before ours (flash trade). 2. The remainder is executed on Chi-X and TOM, while a better price exists on Euronext (sub-optimal order routing) 3. Best offer on TOM was not executed (sub-optimal order routing) 1. Another party s order at 39.91 was executed before ours (flash trade) 2. We then receive worse prices on Euronext compared to what we could have gotten at Chi-X and TOM. (suboptimal order routing) 1. Another party s order at 62,10 and 62,11 was executed before ours (flash trade) 2. We were executed on Euronext for a worse price than what was available on Chi-X. (sub-optimal order routing) 1. Another order to sell at 61,99 was executed before ours (flash trade). 2. We were able to sell further on Chi-X, but instead were executed on Euronext for worse prices. (sub-optimal order routing) 1. We bought on Chi-X at 208.30, while a better price exists on Euronext for 208.25. (sub-optimal order routing) 1. Wij sell on Chi-X at 207,90 while a better price existed on Euronext. (sub-optimal order routing) 2. The bid price on Euronext for 208.00 was executed against another party s order (flash trade). - Page 8 -

Within the annexes, each of the tests outlined above are carried out and presented in great detail. This was facilitated by Bloomberg prints, through which we could determine: 1. The then-current prices for each of the stocks across all trading venues. 2. The precise order of executions against our order across all trading venues. For the first test, we perform a further sub-test, in which we also analyze the exact timestamps (at sub-second granularity) given by TOM s Smart Order Router as well as those from Euronext for executions against our order. This shows that: Our order was split The split order arrives first at TOM, and results in an execution. Between the times at which our order arrives on TOM and on Euronext, a flash trade (as described in the Annex) is executed on Euronext. The possibility of flash trading arises due to the different round-trip times that the TOM Smart Order Router spends sending orders to the different venues. A roundtrip time is the length of time elapsed between the transmission of a signal, to receipt of an acknowledgement. The difference as per the data is 3 milliseconds, which gives market makers plenty of time to initiate a flash trade. The figure below illustrates this point: More details regarding the above can be found in Appendix 2. - Page 9 -

Results and Interpretation: The Flash Boys Scenario The results of the tests clearly show, in almost all cases, that utilizing TOM s Smart Order Router results in sub-optimal execution quality, and specifically that we do not receive the best prices across any of the trading platforms. Further, individual customers of Binck/TOM are quantifiably worse off when utilizing the Smart Order Router. How is this possible? From the analysis of executions against our order, two things are apparent: 1. Our orders are almost always adversely affected by flash trades. This means that trades do occur at the market price at the moment our order is entered, but these trades are not executed against our order. 2. The Smart Order Router divides our order across the various exchanges suboptimally. The results of the tests show that, when utilizing TOM s Smart Order Routing, the execution quality of customers orders are adversely affected by high-frequency traders who perform flash trades. This phenomenon is illustrated in Michael Lewis book Flash Boys. Below, we draw parallels between the events outlined in the book, and the issues we encountered with Binck / TOM. Flash Boys In Flash Boys, Michael Lewis describes how banks / brokers route customer orders to different trading platforms. The primary purpose of this is for them to have the flexibility of choosing a trading venue with the lowest fees. Often, this results in choosing a lowliquidity exchange on which high-frequency traders and market makers are very active. Some of these trading venues are ones in which the market makers themselves have stakes. Other times, the platform can be the dark pool of the broker itself. When a small part of a customer s order is executed on such a venue, the market maker can easily deduce that part of that order has yet to be executed. Furthermore, the market maker can know which stock exchange the broker s Smart Order Router intends to submit the remainder of that order to next. The market maker, who has access to high-speed electronic communication lines, is able to trade on that exchange at the then-current price ahead of the customer s order, leaving the customer adversely affected by way of a worse execution price than they (the customer) would expect. It is the routing facility of the broker, plus the availability of multiple trading venues, that makes it possible for high-speed market makers to anticipate and trade ahead of the customer order flow, resulting in poor execution quality from the point of view of the customer. Flash Boys further asserts that the above is a deliberate strategy of both the brokers and market makers a cooperation that works as follows: Market makers set up electronic trading venues. By offering brokers low fees for executing on these venues, the brokers will naturally route customer orders there primarily. Following the procedure described above, the market maker then deduces information regarding the remainder of these orders, and uses this information to their advantage, by acting faster than the routing facility of the broker. - Page 10 -

Binck/TOM comparison to Flash Boys There is a clear parallel between the concepts illustrated in Flash Boys, and the orderrouting facility offered by Binck/TOM, namely based on the following: 1. The market-maker and/or Broker create a new trading platform, on which the broker can route orders at very low cost relative to the main exchange or other venues. The market maker benefits from the information they can deduce from the order flow. Binck, together with Optiver, are the founders of TOM and are still shareholders in that platform. 2. The broker routes its customer order flow via its Smart Order Router. Binck makes use of TOM s best execution services as its Smart Order Router. 3. Market makers use the information deduced from the brokers customers order flow in order to perform trades at advantageous prices on other exchanges (such as the main exchange) so-called flash trading. This phenomenon is evidenced by the results of the tests we have undertaken: In virtually every case, trades against other orders are executed immediately before our own. 4. The broker consequently enjoys lower fees when routing orders to the electronic trading venue: Binck has praised the efficiency of utilizing TOM in its annual report. 5. The broker benefits from the operations of the trading venue: Binck is a shareholder in TOM. 6. The customer suffers demonstrably and quantifiably poorer execution quality than they otherwise would benefit from without the use of the Smart Order Router. We have quantified the loss incurred via the use of Binck s services in each of our tests - Page 11 -

Conclusion This report has set out to investigate the assertion that the routing facility offered by Binck/TOM is beneficial to customers. Crucial to the service offered by Binck / TOM is the implementation of the Smart Order Router a system that processes the customer s order by sending parts of it to various trading venues. In the context of the tests we have undertaken, we utilized TOM s "best execution services," as the Smart Order Router. In our setup, we sent orders in Euronextlisted products to Binck / TOM, who in turn routed these orders across TOM, Euronext, and Chi-X. Our tests involved sending orders in a number of Euronext-listed shares, with prices slightly worse than the current market price, and quantities larger than the across-venue total available volume. We then compare the quality of the executions we receive from Binck, by quantifying the loss incurred using their facility, relative to what our execution would have been had we traded only on the main exchange. Our results clearly show that utilizing the facility offered by Binck / TOM almost always results in poorer execution quality directly contradicting their claims. Importantly, we show clear evidence of adverse flash trading across trading platforms, which acts purely to the detriment of customer orders through Binck / TOM s Smart Order Router. In particular, Appendix 2 supports this claim with evidence based on timestamps with sub-second granularity from both TOM and Euronext with regards to the execution quality of our first test order. The sub-optimal performance of Binck / TOM s Smart Order Router bears a striking parallel with the ideas espoused in Michael Lewis Flash Boys. Specifically: 1. Binck and market makers have created a new, cheap trading venue, in which Binck is a founder and shareholder. 2. Binck / TOM handle customer orders with the benefit of a Smart Order Router. 3. Market makers anticipate client order flow and execute trade on other platforms. 4. Binck benefits, because it has lower operating costs (routing to cheap platforms). 5. Binck and market makers benefit from their participation in TOM. 6. The main victim of this setup is Binck s customers, who incur substantial costs arising from poor execution quality. All in all, the Smart Order Router facility offered by Binck via TOM s best execution services demonstrably does not yield better execution quality for Binck s customers. For the benefit of our clients, DeGiro will certainly not be making use of the (current version of) TOM s best-execution services. DeGiro transmits orders directly to Euronext because it not only offers the greatest liquidity, but also offers greater liquidity deeper in the levels of the limit order book. The incremental benefit that customers can enjoy, having their orders routed across multiple platforms is, as we have demonstrated, not realized given the current practices of Binck and the costs incurred via adverse flash trading. There is a trend across European markets wherein brokers and market makers increasingly cooperate. DeGiro s position in this environment is one of extreme caution towards utilizing these services, and in particular an aversion to facilities which involve Smart Order Routing. - Page 12 -

Test 1 Date : 18 November 2014 Stock : Solvay (Bloomberg code : SOLB BB Equity) Order : Buy 2000 shares at 113,30 Time : 13:26:28 Buy and Sell prices across multiple trading venues can be seen in the pictures below: 1. Chi-X Start - Page 13 -

2. Tom Start 3. Euronext Start Across these venues, we can buy 1021 shares at a price of 113,25 (103 Chi-X + 200 Tom + 718 Euronext ). - Page 14 -

When we place an order to buy 2000 shares at 113,30, we receive the following execution report from Binck: Solvay Buy 2000 at 113.30 Order Status Quantity Price Date Time Order Placed 2000 113,30 18/11/2014 13:26:32 Partial Execution 50 113,30 18/11/2014 13:26:32 Partial Execution 50 113,30 18/11/2014 13:26:32 Partial Execution 51 113,25 18/11/2014 13:26:32 Partial Execution 50 113,30 18/11/2014 13:26:32 Partial Execution 50 113,30 18/11/2014 13:26:32 Partial Execution 79 113,30 18/11/2014 13:26:32 Partial Execution 200 113,30 18/11/2014 13:26:33 Partial Execution 48 113,30 18/11/2014 13:26:33 Partial Execution 200 113,25 18/11/2014 13:26:33 Partial Execution 51 113,30 18/11/2014 13:26:33 Partial Execution 357 113,30 18/11/2014 13:26:33 Partial Execution 37 113,30 18/11/2014 13:26:33 Partial Execution 52 113,25 18/11/2014 13:26:33 Partial Execution 51 113,30 18/11/2014 13:26:33 Partial Execution 23 113,30 18/11/2014 13:26:33 Partial Execution 72 113,30 18/11/2014 13:26:33 Partial Execution 44 113,30 18/11/2014 13:26:33 Partial Execution 45 113,30 18/11/2014 13:26:33 Partial Execution 22 113,30 18/11/2014 13:26:33 Partial Execution 50 113,30 18/11/2014 13:26:33 Partial Execution 50 113,30 18/11/2014 13:26:33 Partial Execution 30 113,30 18/11/2014 13:26:33 Partial Execution 23 113,30 18/11/2014 13:26:33 Partial Execution 100 113,30 18/11/2014 13:26:33 Partial Execution 100 113,30 18/11/2014 13:26:33 Remainder Executed 115 113,30 18/11/2014 13:26:33 - Page 15 -

The same list of executions reported back from Binck, but this time sorted by exchange and by the order of execution per exchange: Product Exchange Solvay Euronext 2 x 718 @ 113.20-113.25 Buy TOM 200 x 200 @ 113.15-113.25 2000 at 113.30 Chi-X 70 x 103 @ 113.20-113.25 Exchange Order Status Amount Price Date Time Order Book After Transaction Order Placed 2000 113,30 18/11/2014 13:26:32 TOM TOM TOM TOM TOM TOM EUR EUR EUR EUR EUR EUR EUR EUR EUR EUR EUR Chi-X Chi-X Chi-X Chi-X Chi-X Chi-X Chi-X Chi-X Chi-X Partial Execution 200 113,25 18/11/2014 13:26:33 Partial Execution 200 113,30 18/11/2014 13:26:33 Partial Execution 50 113,30 18/11/2014 13:26:32 Partial Execution 50 113,30 18/11/2014 13:26:32 Partial Execution 50 113,30 18/11/2014 13:26:32 Partial Execution 50 113,30 18/11/2014 13:26:32 Partial Execution 79 113,30 18/11/2014 13:26:32 Partial Execution 48 113,30 18/11/2014 13:26:33 Partial Execution 51 113,30 18/11/2014 13:26:33 Partial Execution 357 113,30 18/11/2014 13:26:33 Partial Execution 37 113,30 18/11/2014 13:26:33 Partial Execution 51 113,30 18/11/2014 13:26:33 Partial Execution 23 113,30 18/11/2014 13:26:33 Partial Execution 72 113,30 18/11/2014 13:26:33 Partial Execution 100 113,30 18/11/2014 13:26:33 Partial Execution 100 113,30 18/11/2014 13:26:33 Remainder Executed 115 113,30 18/11/2014 13:26:33 Partial Execution 51 113,25 18/11/2014 13:26:32 Partial Execution 52 113,25 18/11/2014 13:26:33 Partial Execution 44 113,30 18/11/2014 13:26:33 Partial Execution 45 113,30 18/11/2014 13:26:33 Partial Execution 22 113,30 18/11/2014 13:26:33 Partial Execution 50 113,30 18/11/2014 13:26:33 Partial Execution 50 113,30 18/11/2014 13:26:33 Partial Execution 30 113,30 18/11/2014 13:26:33 Partial Execution 23 113,30 18/11/2014 13:26:33???????????????????? 400 x no ask @ 113.20 / 122 x 639 @ 113.25-113.30 122 x 591 @ 113.25-113.30 122 x 540 @ 113.25-113.30 122 x 183 @ 113.25-113.30 122 x 146 @ 113.25-113.30 122 x 95 @ 113.25-113.30 122 x 72 @ 113.25-113.30 122 x 100 @ 113.25-113.30 192 x 100 @ 113.25-113.30?? x?? 192x1277 @ 113.25 113.35 no bid x 51 @ no bid - 113.25 no bid x 264 @ no bid - 113.30 no bid x 220 @ no bid - 113.30 no bid x 175 @ no bid - 113.30 no bid x 153 @ no bid - 113.30 no bid x 103 @ no bid - 113.30 no bid x 53 @ no bid - 113.30 no bid x 23 @ no bid - 113.30 115 x 700 @ 113.20-113.35 - Page 16 -

Notes 1. We received no executions on Euronext against our buy order at the thencurrent price of 113.25. All 718 shares available at that price were executed against another party s order. 2. There appears to be another execution between our first purchase and when the remainder was finally executed. Costs: Flashtrade (718*0.05 ) 35.90 Trading at the wrong price, reason unclear - Total cost 35.90 - Page 17 -

On Chi-X, as expected, we bought 103 shares at 113.25. Then we bought 264 shares at 113.30 us - Page 18 -

On TOM we bought 200 shares at 113.25, as expected. us Followed by buying 400 shares at 113.30. us - Page 19 -

We did not receive any executions against our buy order at 113.25. All 718 stocks were executed against another party s order. other other First trade other - Page 20 -

We receive an execution against our buy order on Euronext, but at 113.30, immediately after the other party s order was executed at 113.25. other Us Us other First Trade - Page 21 -

Our executions appear to be split. Another execution comes in between our executions at 113,30. Our last trade Us Another party s execution between ours For all 12 example trades we would like to direct you to the original document starting from page 14. https://www.degiro.nl/data/pdf/onderzoeksverslag_tom.pdf - Page 22 -

Appendix 2 In this appendix, we present an analysis of the information we have received from TOM and Euronext regarding our first test case the purchase of 2,000 shares of Solvay at a price of 113.30. The essential features are the exact timestamps per line. First, we present two source documents, followed by the analysis. Document from Binck showing order details, executions, and timestamps. Zooming in on the left-hand column showing a clearer view of the timestamps. - Page 23 -

Transactions executed on Euronext - Page 24 -

Analysis of the above Documents We observe the following: The order is sent to TOM at 978 ms, and the acknowledgment is received at 981 ms. The round-trip time is 3 ms, which implies that TOM receives our order at 979.5 ms. The order is sent to Euronext at 974 ms, and the acknowledgment is received at 991 ms. The round-trip time is 17 ms, which implies that Euronext receives our order at 982.5 ms. The above is supported by Euronext s execution report, where we indeed see that our order for 79 shares is executed at 982.5 ms. There is a gap of 3 ms between when the order is received at TOM and when it is received at Euronext (982.5 979.5). This gap is exploited by a fast-acting flash trader transacting on Euronext, with the benefit of knowing the order details from TOM. This is evidenced by the fact that the first transaction, which occured at a favourable price and which we did not get, was executed at 981.2ms. That is almost 2 ms after the arrival of our order on TOM, and 1.3 ms before the arrival of our order on Euronext. It appears that the flash trader was able to fit through the aforementioned 3ms TOM/Euronext gap. DeGiro s unoptimized roundtrip to Euronext ranges between 2 and 3 ms (implying 1 to 1.5 ms one-way). This makes it likely that a flash trader, who has optimized their round trip times, is able to exploit this gap and execute a flash trade in less than 2ms on Euronext. We illustrate the above sequence of events in the following timeline diagram: - Page 25 -