High Frequency Trading 2012

BISE – DISCUSSION

High Frequency Trading

Costs and Benefits in Securities Trading and its Necessityof Regulations

DOI 10.1007/s12599-012-0205-9

The Authors

Prof. Dr. Christoph LattemannSchool of Humanities and SocialSciencesJacobs University Bremen28759 [email protected]

Prof. Dr. Peter Loos (�)IWi at DFKISaarland University66123 Saarbrü[email protected]

Dr. Johannes GomolkaTempelhove Research10117 BerlinGermany

Prof. Dr. Hans-Peter BurghofArne BreuerProf. Dr. Peter GomberMichael KrogmannDr. Joachim NagelRainer RiessProf. Dr. Ryan RiordanDr. Rafael Zajonz

Published online: 2012-03-06

This article is also available in Ger-man in print and via http://www.wirtschaftsinformatik.de: LattemannC, Loos P, Gomolka J, BurghofH-P, Breuer A, Gomber P, Krog-mann M, Nagel J, Riess R, RiordanR, Zajonz R (2012) High Frequen-cy Trading. Kosten und Nutzen imWertpapierhandel und Notwendig-keit der Marktregulierung. WIRT-SCHAFTSINFORMATIK. doi: 10.1007/s11576-012-0311-9.

© Gabler Verlag 2012

1 Introduction

Recent publications reveal that high fre-quency trading (HFT) is responsible for10 to 70 per cent of the order volume instock and derivatives trading (Gomber etal. 2011; Hendershott and Riordan 2011;Zhang 2010). This observation leads toa controversial debate over positive andnegative implications of HFT for the liq-uidity and efficiency of electronic secu-rities markets and over the costs andbenefits of and needs for market reg-ulation. Currently the European Union(EU) is considering the introduction ofa financial transaction tax to curtail theharmful effects of HFT strategies. Theconsideration behind this market pol-icy is based on the assumption that thevery short-term oriented HFT tradingstrategies lead to market frictions. Thiscurrent discourse shows that the argu-ing parties do not homogeneously de-fine HFT. Reasons for this are the pro-ponents’ different but intertwined per-spectives, which lead to new unansweredquestions in numerous subjects of ex-pertise. From a macroeconomic point ofview the question arises if HFT constrainsor supports the allocation function of fi-nancial markets. Capital market researchand information management researchraise questions about the future formof intermediation in securities tradingand the coming architecture of markets,about the HFT’s impact on liquidity andabout price volatility. Financial authori-ties and regulators discuss whether HFThas a stabilizing or destabilizing functionon financial systems and how a futureregulation should be shaped.

This collection of articles shall help todevelop a common definition of HFTand contribute to the ongoing discus-sions. To that end we have collected ar-ticles from representatives of informa-tion systems, business management, theDeutsche Bundesbank and the DeutscheBoerse AG. The following scientists andpractitioners participated in the discus-sion (in alphabetical order):

� Prof. Dr. Hans-Peter Burghof and ArneBreuer, Chair of Business Economics,especially Banking and Financial Ser-vices, University of Hohenheim, Ger-many.

� Prof. Dr. Peter Gomber, Chair of Busi-ness Economics, especially e-Finance,Johann Wolfgang Goethe-Universityof Frankfurt, Germany.

� Dr. Joachim Nagel, Member of theBoard of Directors, and Dr. Rafael Za-jonz, Central Market Analysis, Portfo-lio, Deutsche Bundesbank, Frankfurt,Germany.

� Rainer Riess, Managing Director of theFrankfurter Wertpapierbörse (FWB),and Michael Krogmann, ExecutiveVice President of Xetra Market Devel-opment of Deutsche Börse AG, Frank-furt, Germany.

� Prof. Dr. Ryan Riordan, Institute forInformation Systems and Manage-ment, Karlsruhe Institute of Technol-ogy (KIT), Karlsruhe, Germany.

HFT is a part of algorithmic trading. Go-molka (2011) defines algorithmic trad-ing as the processing and/or execu-tion of trading strategies by the meansof intelligent electronic solution rou-tines (known as algorithms). Thus algo-rithmic trading encompasses computer-supported trading as well as computer-generated sell-side and buy-side markettransactions. Algorithmic trading strate-gies can be both short-term and long-term oriented.

In general, HFT is defined as real-timecomputer-generated decision making infinancial trading, without human inter-ference and based on automatized ordergeneration and order management. HFTencompasses short-term trading strate-gies, which – in extreme cases – operatein the range of microseconds using mini-mal price differences. HFT thus results inminimal profit margins per transactionsand exhibits very short holding periods ofsecurities positions.

However, HFT definitions vary andvarious properties of HFT are notconsistently discussed in the literature.

Business & Information Systems Engineering 2|2012 93

BISE – DISCUSSION

Aldridge’s (2009) definition of HFT hold-ing periods range from milliseconds toone day. Durbin (2010) on the otherhand describes HFT as trading strate-gies, which covers seconds or millisec-onds only. According to Brogaard (2010),HFT is extremely short-term buying orselling with the intention to profit fromminimal price fluctuations.

Further characteristics are often men-tioned but are not always included inHFT definitions, such as the exclusive us-age by professional/institutional investorsin proprietary trading, real-time dataprocessing and direct market access (Da-corogna et al. 2001). Another controver-sial issue is the avoidance of overnightrisk (Aldridge 2009). Other definitionsunderline the role of HFT as financialintermediary (Hendershott and Riordan2011) or try to find differences amongthe implemented trading strategies (Ye2011).

On the basis of the broad HFT defini-tion given before the authors in this arti-cle will shed light on the following ques-tions: (1) How does HFT influence theliquidity and efficiency of electronic secu-rities markets? (2) What are the costs andbenefits of, and what are the needs for aHFT regulation?

Peter Gomber analyzes HFT from amarket microstructure perspective, andfinds HFT to be a central element of thevalue creation chain in securities trad-ing. As part of the value creation chain,HFT contributes to increased efficiencyand reduced explicit and implicit transac-tion costs. In his eyes, regulation of HFTcould lead to dramatic changes in marketbehavior, while an inappropriate regula-tion might even be counterproductive formarket quality. Gomber sees the prob-lems for profound research on HFT in thelack of data available for empirical stud-ies. Again this leads to adverse effects indiscussions of the topic in the public, inthe media, and with regulators.

Ryan Riordan also looks at HFT fromthe perspective of market microstructureand interprets HFT as one form of tech-nological financial intermediation whichcontributes to the efficiency of opera-tions in exchange trading. In his eyes,HFT plays an important role in the pro-cess of price formation and influences thesize of transaction costs in securities trad-ing. According to him, one cannot yetsay whether HFT will have a positive ora negative impact on the capital markets.However, he sees major advantages in a

highly technologized market. It is no al-ternative for him to turn back the wheelsand return to a backward oriented, artifi-cially slowed, regulated trading, which isbased on human intermediation.

Rainer Riess and Michael Krogmanndescribe HFT as the highest evolution-ary level of securities trading. In theiropinion HFT leads to faster processingof information, to an increase in liq-uidity, and thus added values for theoverall economy. The authors describehow HFT is currently technically real-ized and integrated into trading opera-tions at the exchange, and deduct theirarguments accordingly. From the point ofview of Deutsche Börse, HFT is mainlyused by institutional investors in pro-prietary trading and focuses on highlyliquid stocks. The authors correlate therise of HFT with a continuous improve-ment of the electronic trading system XE-TRA, which – from the point of viewof Deutsche Börse – benefits all marketparticipants in the same way. In the eyesof Riess and Krogmann, a future regula-tion of HFT should primarily focus onequal chances of competition in the EU-area, in order to create “a level playingfield“. From the point of view of DeutscheBörse, it is necessary not only to imple-ment security mechanisms on the side ofexchanges but also with HFT-firms.

Arne Breuer and Hans-Peter Burghofalso recognize that, due to HFT, infor-mation can be processed more perfectlyand faster than ever before. They look atthe topic from the perspective of finan-cial economics. This point of view leadsthem to believe that more and faster in-formation does not necessarily lead toa correct determination of the intrin-sic value of financial instruments. RatherHFT processes short-term information,which primarily is made of short-termvolume and short-term time series data,and thus does not contribute to the eval-uation of the intrinsic values. The au-thors vote for a stricter regulation of HFT.However, before this can be done, moreanalyses should be conducted. For this,more data are necessary.

Finally, Joachim Nagel und Rafael Za-jonz argue from the perspective of reg-ulators. A blanket judgment on HFT isfrom the regulators’ point of view nei-ther adequate nor would it lead to im-provements of the regulatory frameworkregarding transparency, stability, and ef-ficiency. The impact of HFT on the effi-ciency of securities trading is – due to theabsence of a scientific discussion – still

unclear for the regulators. The possibilityto destabilize the market due to HFT involatile market situations is regarded ascritical but should be looked into in de-tail. From the point of view of the authors“market friendly” strategies exist, a factwhich can be judged positively. But thereare also ”unfriendly strategies“, which –from their perspective – can be catego-rized as potentially harmful. In the cen-ter of their article, the authors formu-late the wish that this complex topic maybe discussed more intensely by the sci-entific community in the future, in orderto better understand which fundamental,regulatory measures should be applied toHFT.

If you would like to commenton this topic or another article ofthe journal Business & InformationSystems Engineering, please sendyour contribution (max. 2 pages) tothe editor-in-chief, Prof. Hans Ul-rich Buhl, University of Augsburg,[email protected].

Prof. Dr. Christoph LattemannSchool of Humanities and Social

SciencesJacobs University Bremen

Prof. Dr. Peter LoosIWi at DFKI

Saarland UniversityDr. Johannes Gomolka

Tempelhove Research

2 High Frequency TradingRegulation Required at aReasonable Level

It is uncommon for a specific subjectin the field of securities trading and IT-innovation to draw as much public atten-tion as high frequency trading (HFT) hasbeen doing in recent months. Merely aspecial field for a small group of expertsprior to 2010, it is now a frequent partof the general news coverage. Against thebackground of the recent debt crisis, thecurrent volatility and market turmoil aswell as the “US Flash Crash” on May 6,2010 lead to this extreme attention. Sev-eral parties attempt to exert pressure onpolitics and regulation by making HFTresponsible for that crisis and the highmarket volatility. In reaction to the afore-mentioned incidents and to the subse-quent public discussions, the regulatory

94 Business & Information Systems Engineering 2|2012

BISE – DISCUSSION

authorities of international financial cen-ters have debated the adoption of vari-ous regulatory measures and now pro-pose regulatory procedures, which cur-rently substantiate especially in Europeand will presumably be approved in 2012in the context of the revision of the Mar-kets in Financial Instruments Directive(MiFID).

Basically, the trading strategies basedon HFT can be subdivided into ac-tive and passive ones. Whereas passivestrategies provide other market partici-pants with trading opportunities in termsof quotes and limit orders (e.g. elec-tronic market making), active strate-gies primarily attempt to exploit imbal-ances of asset prices in fragmented mar-kets (e.g. primary market and multilat-eral trading facilities), discrepancies invaluation between different asset classes(e.g. between derivatives and their un-derlyings) or deviances of current as-set valuations compared to historicalcorrelations (e.g. in the so-called pairstrading) immediately after the emer-gence of these trading/arbitrage opportu-nities.

The emerging academic literature,which analyzes the effects of HFT basedstrategies on market quality, showsmostly positive impact (for a system-atic outline of academic research con-cerning HFT see Gomber et al. 2011).Regarding price discovery, liquidity andvolatility, most studies discover positiveeffects of HFT. Only a few publicationsindicate that HFT can increase the ad-verse selection problem under specificcircumstances, and in the case of the “USFlash Crash” another survey (Kirilenkoet al. 2011) reveals that HFT can increasevolatility.

The growing market efficiency and areduction of explicit and implicit trans-action costs triggered by HFT is an ob-vious issue particularly for those mar-ket participants who used to capitalize onintermediary services and broad bid/askspreads in a formerly less efficient andless liquid trading environment. In con-trast to off-exchange trading via inter-nalization and so-called dark pools, i.e.non-transparent execution venues, HFTmarket-making strategies on lit marketsface relevant adverse selection costs asthey provide liquidity on the marketwithout knowing their counterparties.Within their internalization systems anddark pools in the OTC field, banks andbrokers are aware of their counterparties’

identities and can benefit from this infor-mation. Contrary to this, HFTs in lit mar-kets are not aware of the toxicity of theircounterparts and are – analogous to mar-ket makers – exposed to the problem ofadverse selection.

Inappropriate regulation of HFT basedstrategies or an impact on HFT businessmodels due to excessive burdens mightturn out to be counterproductive andlead to unforeseeable consequences forthe quality of markets. However, abusivestrategies have to be combated effectivelyby the regulators. Particularly the analy-sis of the “US Flash Crash” with its dis-cussed solution approaches can hardlybe transferred to the European situation,since the issues related to the “US FlashCrash” primarily result from the US mar-ket structure. In Europe, where a moreflexible best execution regime is imple-mented and a share-by-share volatilitysafeguard regime has been in place fortwo decades, no market quality problemsrelated to HFT have been documented sofar. Therefore, a European approach tothe subject matter is required, and Eu-rope should be cautious about address-ing and fixing a problem that exists in adifferent market structure and thus creat-ing risks for market efficiency and marketquality.

Any regulatory interventions in Europeshould try to preserve the benefits of HFTwhile mitigating the risks as far as pos-sible by assuring that (i) HFT firms areable to provide documentation on theiralgorithms upon authorities’ request andto conduct back-testing, (ii) markets arecapable of handling peaks in trading ac-tivity and apply safeguards to react totechnical issues of their members’ algo-rithms, (iii) a diversity of trading strate-gies prevails to prevent systemic risks,(iv) co-location and proximity servicesare implemented on a level playing field,(v) regulators have a complete overviewof the possible systemic risks which couldbe triggered by HFT, and have employ-ees who have the knowledge and the toolsto assess the impact of the trading al-gorithms on market quality and the as-sociated risks. Furthermore, it is crucialthat market places in a fragmented envi-ronment develop coordinated safeguardsund circuit breakers, which mirror theHFT reality and enable all market partic-ipants to react adequately even in marketstress.

Regulatory proposals demanding con-tinuous liquidity provision by HFT in thesense of market marking obligations or

minimum quote lifetimes miss the markand are not suitable to improve marketstability or market integrity. They rathercontribute to a decrease in market qualityand higher transaction costs.

At first sight, demanding obligationsfor HFTs to provide quotes seems an ap-propriate measure to tackle the problemof a sudden liquidity withdrawal. How-ever, it is highly doubtful whether anyrule can force market makers to buy inthe face of overwhelming selling pres-sure. In such a situation they might rathertake the risk of being fined for not fulfill-ing their obligations. Many HFT strate-gies are characterized by rapid closingof built-up positions to minimize risk.Hence, an obligation to provide liquidityand thereby risk capital is in sharp con-trast to many HFT business models. Dueto the significant regulatory costs thoseobligations would potentially lead to aretreat from the market and thus to anotable loss of liquidity.

Also a minimum order lifetime, whichat first glance appears to be useful toavoid fast order submissions and im-mediate cancellations, would lead to asignificant change in market behavior.Market participants are then no longerable to react quickly and adequately tomarket-exogenous information (e.g. ad-hoc news) and the necessity to keep anorder in the order book presents a freeoption for other market participants. Be-sides, the existence of minimum orderlifetimes would lead to an implementa-tion of trading strategies capitalizing onthe “lock in” of orders. HFT would an-ticipate the accompanied risks and costsand attempt to compensate these costswith higher spreads, which again wouldhave negative effects on market quality.In this debate it should not be neglectedthat speed is the key tool for HFTs’ riskmanagement.

HFT is an important factor in mar-kets that are driven by sophisticated tech-nology on all layers of the trading valuechain. However, discussions on this topicoften lack sufficient and precise informa-tion. A remarkable gap between the re-sults of academic research on HFT andits perceived impact on markets in public,media and regulatory discussions (Euro-pean Commission 2010) can be observed.Here, the provision of granular and reli-able data by the industry can assist em-pirical research at the interface of financeand IS to provide important contribu-tions to a reasonable regulation of HFT.

Business & Information Systems Engineering 2|2012 95

BISE – DISCUSSION

This regulation should eventually mini-mize the inherent risks of the technol-ogy in question without hindering theindisputably existing positive effects formarket quality.

Prof. Dr. Peter GomberUniversity of Frankfurt

E-Finance Lab

3 High Frequency Trading (HFT) –A New Intermediary

Financial markets require intermediariesto provide liquidity and immediacy forother participants. These intermediaries,often called market makers or special-ists, were often afforded special statusand located on the trading floor, or closeto the trading mechanism of exchanges.The automation of financial markets hasincreased their trading capacity and in-termediaries have expanded their use oftechnology. This has resulted in a reducedrole for traditional human market mak-ers and led to the rise of a new intermedi-ary, referred to as high frequency traders(HFTs).

This development has been made pos-sible by the technological innovations inrecent years. HFT strategies usually makeuse of the high speed technologies tobuild up and unwind positions withinmilli- and microseconds. Prerequisitesfor this development were the reduc-tion of system latency and the increaseof computing power and data process-ing capabilities of computers. Next to thelarge investments in HFT, exchanges havealso invested large amounts of money intheir IT infrastructure. For example, thecosts of a high-speed connection betweenChicago and New York are estimatedaround $200,000 per mile (Forbes 2010).The question remains whether these in-vestments are justified with regard to theincrease of overall market quality andwelfare that results from higher HFTactivity on the market.

Like traditional intermediaries HFTshold little inventory, have short hold-ing periods, and trade often. Unlike tra-ditional intermediaries, however, HFTsare not granted preferential access to themarket not available to others and theyemploy advanced and innovative tech-nology to intermediate trading. With-out such privileges, there is no clear ba-sis for imposing the traditional obliga-tions of market makers on HFT. The sub-stantial, largely negative media coverage

of HFT and the so called “flash crash”on May 6, 2010 raise significant interestand concerns about the role HFT play inthe stability and price efficiency of finan-cial markets. The predominantly negativecoverage seems mostly unfounded.

Overall, HFTs’ impact is similar tothat of other intermediaries and specu-lators. Speculators can improve price ef-ficiency by obtaining more informationon prices and by trading against pric-ing errors. Manipulative strategies andpredatory trading could decrease priceefficiency. Reducing pricing errors im-proves the efficiency of prices. HFTs’ in-formational advantage, which is drivenby the technology they employ, is short-term. It is unclear whether or not thisshort-term information and intraday re-ductions of pricing errors facilitate betterfinancial decisions and resource alloca-tions by firms and investors. If the short-term information – that HFTs price in –would not otherwise become public mi-croseconds later, HFT clearly plays an im-portant role (Hendershott and Riordan2011). It would be an important positiverole of smaller pricing errors if these cor-responded to lower implicit transactioncosts by long-term investors.

One important point left unaddressedthus far is whether or not HFTs engage inmanipulative or predatory trading. Theiruse of technology may allow HFTs to ma-nipulate prices at speeds that are unde-tectable by slower traders. A manipula-tive strategy might be the ignition of aprice movement in one direction only inorder to trade on the opposite side of themarket as proposed by the SEC (2010)and therefore cause significant pricingerrors. As is frequently done, one canargue whether the underlying problemof possible manipulation lies with themanipulator or the market participantwho is manipulated. In the SEC exam-ple, the passive manipulation could notsucceed if there were no price matching.The manipulation stories could be testedwith more detailed data identifying eachmarket participant’s orders, trading, andpositions in all markets.

Despite the strong evidence of the pos-itive role of HFT for the efficiency ofprice determination and trading costs(Hendershott et al. 2011; Brogaard 2010;Zhang and Riordan 2011), regulators andthe media are certain that they must beregulated. It is, however, unclear and alsodebatable how we should regulate HFT.Assuming that some, or most, of their

activities contribute positively to liquid-ity and price efficiency, which parts oftheir trading should we regulate? Thereare controversially discussed suggestionsto restrict HFTs’ mostly passive trading orto enforce a minimum order life on limitorders. Restricting HFTs’ ability to tradeactively necessarily impedes their abilityto manage the risks associated with in-termediation. This may lead to less in-termediation and lower liquidity. Impos-ing minimum order lives on limit or-ders may also negatively impact HFTs’ability to manage trading risks duringvolatile market periods that existed be-fore HFT dominated the equity market.Finally, the discussions of US and Euro-pean regulation should take into accountspecific differences of both markets. De-spite the high market fragmentation, theEuropean market has maintained a com-parably high degree of efficiency. This isalso due to the help of HFTs. They makeuse of arbitrage strategies to dissolve ex-isting price deviations within secondswhich results in an interconnectedness ofEuropean markets.

A final point is a more general oneon technology investments. HFTs mustmake a large and long-term investment intechnology, both hardware and software.This investment in technology seems tohave to paid-off both for HFTs and theequity markets. If regulation were tonegatively impact the returns on invest-ments in HFT technologies by reducingthe profitability of intermediation, fewerfirms will be willing to invest in thesetechnologies. This may lead to a situ-ation in which one or two highly spe-cialized firms dominate intermediation,which ultimately leads to less competi-tion, lower liquidity and reduced price-efficiency. Competition, ease of marketentry and the use of specialized and in-novative technology seem to be the bestguarantors of market stability.

It is hard to imagine a situation inwhich HFTs are able to artificially ma-nipulate prices for longer periods oftime given the intense competition otherHFTs. HFTs are one type of intermediary.When thinking about the role HFT playsin markets it is natural to try to comparethe new market structure to the previ-ous market structure. Some primary dif-ferences are that there is free entry intoHFT, HFTs do not have a designated rolewith special privileges, and HFTs do nothave special obligations. When consider-ing the optimal industrial organization ofthe intermediation sector, which includes

96 Business & Information Systems Engineering 2|2012

BISE – DISCUSSION

regulation, market structure, technologyand incumbency, HFT more closely re-sembles a highly competitive environ-ment than traditional market structures.A central question is whether there werebenefits of the more highly regulated andless technology intensive intermediationsector which outweigh the costs of lowerinnovation and higher entry costs typi-cally associated with regulation. The an-swer to this question seems thus far to bea resounding “no”.

Prof. Dr. Ryan RiordanKarlsruhe Institute of Technology

4 High Frequency Trading – AnExchange Operator’s Perspective

4.1 High Frequency Trading – Myth andReality

On 2010-09-30, the U.S. Securities andExchange Commission (SEC) and theCommodity Futures Trading Commis-sion (CFTC) (2010) issued a joint reportshowing that the so-called “flash crash”,a sequence of events which made pricesplunge throughout the US stock mar-ket, was caused by an incorrectly pro-grammed trading algorithm of a tradi-tional investment company which didnot use high frequency trading (HFT).Nevertheless, HFT has gained massivepublic attention ever since. The news me-dia, as well scientists and regulatory au-thorities, are busy discussing and analyz-ing the effect of HFT on the global capi-tal markets. While the public perceptionof HFT is largely critical – and driven byheadlines demanding a HFT ban or, atleast, strict regulation – scientific analysiscomes to rather different conclusions (seeGomber’s discussion above). Accordingto Brogaard’s (2010) study of HFT, blam-ing HFT for the US flash crash is notthe only popular fallacy regarding therole of HFT in securities trading. Bro-gaard’s analysis of NASDAQ data showedthat for 65% of the time HFT accountedfor the best bid and ask quotes. Also,Broogard found no evidence suggestingthat HFT firms systematically engage inmarket abuse, e.g. by illegally taking ad-vantage of information about client or-ders, the so-called “front running”. SinceHFT firms are proprietary traders, theydo not have any clients. Generally, sci-entific analysis did not find a correla-tion between HFT and market abuse.

The Netherlands Authority for the Finan-cial Markets (AFM 2010) considers HFTas a legitimate trading method which isnot market abusive under normal cir-cumstances. According to Gomber, aca-demic papers mostly could not find ev-idence for negative effects of HFT onmarket quality. Moreover, the Germany-based Karlsruhe Institute of Technology(KIT) concluded their study based onanalysis of NASDAQ data with the find-ing that HFT even worked as a bufferagainst plunging stock prices during thecrisis years 2008 and 2009 (Zhang andRiordan 2011).

4.2 Insights of an Exchange Operator

We live in a technology-driven society,continuously striving to further improveand advance the achievement potentialof our economy as well as of nearlyevery aspect in our everyday life: cananyone imagine a commercial flight to-day without the aid of an autopilot, ormodern microsurgery without robotics?These advancements are by no meansends in themselves but serve a greatergood. Just the same goes for the everincreasing speed in securities trading –a development which leads to continu-ously improving general market qualityand also to more efficient risk manage-ment for every market participant. Thefaster the market data transmission, thefaster investors are able to adapt to ongo-ing market developments. This does notonly have a very positive effect on thesafety in securities trading but also ontransaction cost: faster trading leads totighter spreads and, therefore, to higherliquidity. The implicit transaction costsof every securities trade are determinedmainly by liquidity and account for upto 80 percent of the overall transactioncosts, while the explicit transaction costs– commissions, fees, taxes – are of minorsignificance. With this in mind, DeutscheBörse started long before the advent ofHFT to improve the trading infrastruc-ture of its electronic trading platform Xe-tra, especially in view of ever decreas-ing systemic latency. At the same time,Deutsche Börse further developed the se-curity mechanisms and technologies re-spectively adapted them to the increas-ing demands of a more and more sophis-ticated and faster trading system, one ofthem being the very effective instrumentof the volatility interruption, introducedin 1999. This security mechanism is usedin extremely volatile market phases and

leads to higher price stability: wheneveran indicative price is outside the pricerange – which is pre-defined for every se-curity traded on Xetra – a volatility in-terruption will be initiated around thereference price.

While continuously advancing thetechnical infrastructure, Deutsche Börseexpanded its offer of individually se-lectable bandwidths for market par-ticipants connected to Xetra from512 Kbit/sec up to 2 Mbit/sec for theirValues API interfaces. In 2008, for Xe-tra members requiring even faster mar-ket data transmission and more orderbook depth, an additional interface witha bandwidth of 1 Gbit/sec was imple-mented, called Enhanced Broadcast So-lution respectively Enhanced TransactionSolution. Today, bandwidths of up to 10GBit/sec are available. With the intro-duction of the so-called “non-persistent”orders in 2009, Deutsche Börse furtherenabled Xetra members to optimize theirresponse times to price changes thanksto even faster data processing. “Non-persistent” orders are not saved in ex-change systems and are thus designed notbe executed after volatility interruptions.

In late 2011 Deutsche Börse comple-mented its connectivity portfolio withthe FIX (Financial Information Ex-change) gateway. Market participants us-ing this protocol now are able to connectto Xetra far more easily.

However, there was one latency factorleft that even the most sophisticated tech-nology could not overcome: the propaga-tion delay due to physical distance. Forevery 100 km which a market partici-pant’s trading engine and the trading sys-tem of Xetra are physically apart, transac-tion latency increases by 1 msec approxi-mately. This could mean a true competi-tive disadvantage for market participantsrelying on ultra low latency. DeutscheBörse addressed this growing market de-mand by introducing its proximity ser-vices in 2006. By placing the trading en-gine of distant Xetra members not onlyvirtually but physically close to the ex-change back end – a process called co-location – the travel time of the mar-ket data could be drastically reduced. To-day, 141 Xetra members take advantageof Deutsche Börse’s co-location offer.

Thanks to a continuously perfectedtrading infrastructure and the intro-duction of proximity services, DeutscheBörse has not only remained competitiveon an international level but has also pre-pared Xetra optimally for the needs of

Business & Information Systems Engineering 2|2012 97

BISE – DISCUSSION

HFT firms. Over the last few years, sys-temic latency on Xetra has been furtherreduced notwithstanding a dramatic in-crease of technical transactions – an ad-vantage to all market participants alike:a fair, equal access to Xetra and the pre-and post-trade transparency characteris-tic of a regulated exchange make surethat every investor enjoys all advantagesDeutsche Börse’s trading platform has tooffer.

While being a minority, HFT firmsnevertheless play an important role inimproving the order book quality on Xe-tra, e.g. by bundling the very hetero-genic order flow. There are three orga-nized forms of HFT on Xetra: the propri-etary trading of investment firms, hedgefunds, and proprietary trading compa-nies. Two types of trading prevail: first ofall, the so-called electronic liquidity pro-vision. In this case, HFT firms act as vol-untary market makers, adding liquidityto a multitude of securities. The secondtype of HFT on Xetra is called statisticalarbitrage which leads to improved pricediscovery. Both types of HFT accountfor tighter spreads and, ultimately, im-proved market efficiency on Xetra. So far,Deutsche Börse could find no evidence ofHFT having lead to destabilizing marketsduring periods of market turmoil, e.g. bystrengthening trends. During the highlyvolatile market phase in August 2011, thetrading volume on Xetra increased tem-porarily to 107 million transactions onone single day. Despite up to 30 volatil-ity interruptions, the average transactionprocessing took only 0.4 msec longerthan usual. System availability was guar-anteed at all times, Xetra members didnot have to face any restrictions, let alonesystem failure. Deutsche Börse’s marketsecurity mechanisms made sure that alltrading activities could be executed prop-erly and continuously while price stabil-ity was guaranteed even during marketturmoil.

Thus, Deutsche Börse succeeded in ad-vancing the Xetra infrastructure in termsof continuously decreasing systemic la-tency and, at the same time, met thepermanently increasing needs regardingsafety and speed of its trading systemeven before the term HFT came up.

4.3 Regulatory Recommendations

Within a national economy it is the ex-plicit function of a securities exchange tofacilitate the most efficient employment

of capital, ensuring best possible corpo-rate financing and re-financing. HFT, asit is today, supports faster processing ofeconomically relevant data and leads tohigher liquidity in the trading of com-pany shares. Thanks to a stable, high-performance trading system, DeutscheBörse was able to integrate HFT success-fully and to use the positive effects ofHFT to improve overall market quality.This would not have been possible with-out Deutsche Börse’s principle of equalaccess and a fair set of rules applying toevery market participant trading on Xe-tra alike. From a regulatory perspective– and keeping MiFID’s ultimate goal ofcreating an EU-wide “level playing field”in mind – comprehensive rules regard-ing HFT definitely would make sense.Therefore, Deutsche Börse supports allmeasures to enhance transparency, e.g.the complete registration of all marketparticipants and a full recording of alltheir trading activities – traditional trad-ing and HFT alike. The Deutsche Börse(2011) has come to the conclusion thatregulatory intervention in HFT must nothurt the proven positive effect on mar-ket quality HFT has to offer. In particu-lar, the variety of HFT strategies shouldbe preserved, as systemic risk should beprevented. To achieve these goals, HFTfirms themselves may have to implementsecurity mechanisms – just as exchangeoperators as Deutsche Börse already have.

Whichever regulatory rules may be im-plemented in the end, the regulators willhave to make sure that these rules applyto every European market and to everymarket participant in Europe to the verysame extent.

Rainer RiessMichael KrogmannDeutsche Börse AG

5 Paradigm Change ThroughAlgorithmic Trading

5.1 Introduction

Algorithmic trading nowadays often ac-counts for more than half of trade vol-ume and order volume at large stockexchanges. Its net effects are generallyfound positive by researchers. Only fewvoices from the scientific community –more, however, from traders – point outnegative effects of algorithmic trading. Anotable difference lies between empirical

findings – that usually find positive ef-fects – on the one hand, and some theo-retical works and especially the sentimentof traders, who often express their frus-tration about their computerized coun-terparts, on the other hand.

5.2 Availability of Data

Most scientific studies about algorith-mic trading share one fundamental prob-lem: data about algorithmic trading arescarce. As one of the few stock ex-changes, Deutsche Börse had for sometime quite reliable data on algorith-mic trading. Their “Automated TradingProgram“ (ATP), which was in effectfrom 2007 to early 2009, enabled themto distinguish between algorithmic or-ders and human ones (Deutsche Börse2009). Hendershott and Riordan (2011),Gsell (2009), Groth (2009), and Maurerand Schäfer (2011) analyze such datasetswhich contain flags for orders placedwithin the ATP environment. Their re-search questions differ, but they all moreor less conclude that the overall effect ofalgorithmic trading is positive.

A fundamental critique of such analy-ses is that algorithms usually work wellin “normal” markets and then show theoften-found positive effects. The modelsthat algorithms base on are abstractionsof reality and must fail to reflect it in itsentirety. If a market situation is not partof the possibility space of the model, sev-eral options are possible: The algorithmhalts trading and waits until the market is“normal” again, thereby facing the risk togenerate possibly considerable losses. An-other option is to continue trading usingthe usual model, thus failing to trade op-timally and possibly worsening the situa-tion. Since the flash crash on May 6, 2010,there have been repeated miniature flashcrashes that did not affect the whole mar-ket but only individual stocks. For bothphenomena, algorithms are blamed to bethe cause of the market irregularities.

However, an effective approach to reg-ulation should base on well-establishedresults. A lot of work has to be donehere. Above all, the insufficient availabil-ity of appropriate data confines scien-tific progress. The deduction of the ef-fect of algorithmic trading on the mar-ket from anonymous order book data canonly be very rudimentary. In our currentwork, we attempt to find a way to an-alyze algorithmic trading activity whilstonly using anonymous order book data

98 Business & Information Systems Engineering 2|2012

BISE – DISCUSSION

(Breuer and Burghof 2011). A manda-tory flagging of algorithmic orders wouldbe desirable. Only then would it be pos-sible to independently analyze algorith-mic trading from many points of viewand estimate the effect on the market.The restrictive handling of historic ATPdata by Deutsche Börse does not buildconfidence but could increase the prob-ability that the sentiment towards AT isinfluenced by irrational fears.

5.3 Information Efficiency

Recent studies (Hendershott and Riordan2011; Gsell 2009; Groth 2009; Maurerand Schäfer 2011) analyze rather short-term aspects of market microstructure inan AT environment. Indeed, its existencealters behavioral incentives of other mar-ket participants fundamentally and in thelong run. It is apparent that algorithmsprocess new information ever faster and– assuming normal market conditions –probably calculate its price impact betterthan humans. It is still to be seen, how-ever, how accurate trading algorithmsprocess information without slow humantraders monitoring them. Sometimes, thesuperfast processing of news can be un-desirable. An example for this is the newsabout the bankruptcy of United Airways.The airline’s stock price plummeted un-til it became clear that the news was al-ready a couple months old. Because thepossibility to extract yields from new in-formation has a very short and decreas-ing half-life, systems tend to react hastilyand without challenging the information.Especially in delicate market situations,rumors can develop a destructive power.

The effect that is likely to be mostimportant has however escaped scien-tific analysis so far. Capital markets are ahighly efficient instrument of capital al-location, especially because a large num-ber of actors feed information into theprice via their trading activity. This in-formation comes from various sources;it may be obtained haphazardly or withsome effort. Algorithmic trading uncov-ers trade activity which is caused by thatinformation and uses this knowledge topocket a considerable part of the infor-mation yield. The better these algorithmswork, the less money the informed per-son will make out of this information. Inthe long run, this could mean that thecostly generation of information turnsunprofitable, and in an extreme case eventhe trade based on incidentally obtainedinformation does not pay anymore.

In such a hypothetical market, ever lessinformation is traded ever more perfectlyand faster. The market draws nearer andnearer the weak form of market efficiency(Fama 1970) or eventually even the semi-strong form of market efficiency. At thesame time, it moves away from the strongform of market efficiency, because the in-centive to feed information into the mar-ket becomes considerably less powerful.It is this very effect that traders witnesswhen they trade against algorithms. Theyknow that information-based strategiesare detected rapidly and thwarted by ap-propriate front-running strategies (Biaiset al. 2010; Cvitanic and Kirilenko 2010).Surly, there is still a need for theoreticalas well as empirical analysis here, becausedue to these thoughts, the usefulness ofalgorithmic trading is subject to scrutiny.

5.4 Regulation and RegulatoryInstruments

Regulatory considerations have to distin-guish between the different types of algo-rithms. Limit orders which are bogus or-ders or part of quote-stuffing techniqueshave to be considered under the light oflaws against market manipulation (e.g.,§20a (1) No. 2 of the German SecuritiesTrading Act [WpHG]). Other strategiesimprove the price quality by arbitragingprices and equalizing them across differ-ent trading venues. Because of the mar-ket power of algorithms, there is the riskthat overly mechanic thinking and potentalgorithms may perturb the price forma-tion process. Naturally it would be de-sirable to capture the positive effects ofalgorithmic trading and to dampen thepotentially negative ones. There may bemore than one way to reach this aim.

A simple ban of algorithmic trading, assometimes demanded by certain politi-cal circles, cannot serve to reach this dif-ficult aim. This would mean to also de-stroy many preferable effects of algorith-mic trading. Of course, a distinction ofalgorithmic and “normal” trading is noteasy. And certainly market participantswould program algorithms that operatein the gray area to hide their true nature.

Currently, regulatory bodies are dis-cussing possible means (Dombert 2011).The often contemplated lower boundaryfor limit order lifetimes is regarded scep-tically. The comprehensible reason is thatan efficient risk management of orderswould be drastically complicated – es-pecially, but not exclusively, in volatile

markets. Dombert (2011) proposes an al-ternative that is worth discussing. Withan order-transaction-ratio, the numberof orders divided by the number of trans-actions would have to remain above someexogenous constant.

In our view, a European regulatoryframework is desirable that defines theplayground for all market participants.Within this framework, it should be leftto the trading venues how they wish totreat algorithmic trading in the contextof their business model. Then it wouldbe up to them if they wanted to attractalgorithmic trading or to limit it in spe-cific market conditions. Such a “menu-approach” leaves it in essence to the in-dividual trader if he or she wishes toface the competition from algorithmswith all their positive and negative ef-fects or evade them by trading on tradingvenues with appropriate restrictions thatapply always or under specific marketconditions.

5.5 Conclusion

As long as algorithms operate in thedark, there is a profound uncertaintyabout the effect of their activities. There-fore, algorithmic trading is partly in con-tradiction to fundamental principles ofstock exchanges: bringing buyers andsellers together in a transparent manner.On stock exchanges, trust is paramount.The opacity of algorithmic trading –as comprehensible it may be from thepoint of view of their operators – un-dermines this principle. Currently, thereis no level playing field. However, itis equally important to enable techni-cal progress, which algorithmic tradingwith its high-quality information pro-cessing definitely is. An improved avail-ability of data and associated scientificresearch can help to develop reasonableregulatory frameworks for algorithmictrading. With the increasing importanceof this way of trading in mind, there isless and less reason to doubt that the im-plementation of appropriate regulatoryframeworks should have a high priority.

Arne BreuerProf. Dr. Hans-Peter Burghof

University Hohenheim

6 High Frequency Trading –A Central Bank View

The capital markets are currently at animportant juncture in their development.

Business & Information Systems Engineering 2|2012 99

BISE – DISCUSSION

Roughly half of all stock and foreign ex-change trades conducted on the majorexchanges are no longer initiated by hu-man traders; instead, they are the prod-uct of computer algorithms that are ableto analyze large volumes of data and ini-tiate hundreds of orders in fractions ofa second. Humans are increasingly beingeliminated from the immediate decision-making process relating to the sale andpurchase of assets and being replaced bysoftware programs.

The speed with which orders are exe-cuted has become to be the most impor-tant factor and is now measured in milli-and microseconds. New practices such as“co-location” or “quote stuffing” – plac-ing huge quantities of buy or sell orderswhich the instigator intends to cancel al-most immediately before they are exe-cuted – have become important instru-ments in the battle for the most rapid or-der execution. Fundamental data on thevalue of the respective securities or cur-rencies are of no, or only subordinate,importance for HFT algorithms.

In HFT, positions are usually held forbetween a number of milliseconds andseveral hours. In today’s high-speed mar-kets, the scales are no longer tipped infavor of the investor who is best able toassess the true value of an asset, but ofthe investor able to trade fastest. Trueinvestments are becoming increasinglyrare.

Since the “flash crash” of May 6, 2010(a roughly 15-minute phase of unusualand irrational volatility on the New YorkStock Exchange), HFT has been calledto the attention not only of the generalpublic but also of regulators and centralbanks.

Numerous observers regard HFT as anew technical means of executing exist-ing trading strategy rather than a strat-egy in its own right. Advantages in termsof speed have, they say, always been anessential component of many successfultrading strategies. Seen from this per-spective, HFT is not a completely newphenomenon, but rather a technical evo-lution of the securities markets. HFTshould be regarded merely as an overar-ching term covering a multitude of differ-ent fields of use. Among the many tactics,several of the most important are basedon providing liquidity in stock markettrading (market making). Others can beincluded under the category “statisticalarbitrage” and use algorithms to swiftlyidentify and exploit profitable trading

opportunities based on price data. Oth-ers belong to a category known as liq-uidity detection, in which traders try toseek out hidden large orders in orderbooks. Many critics term this “preda-tory trading”, and it is suspected of beingunfair and potentially damaging to themarket.

Against this complex background, anyassessment of HFT and all discussionrelating to potential regulation should,where possible, be limited to the un-derlying HFT strategy. From a centralbank perspective, a sweeping judgmenton HFT is therefore neither appropri-ate, nor would it serve to improve theregulatory framework for transparency,stability and efficiency. That means thatboth the advantages and disadvantages ofHFT need to be evaluated very specifi-cally. Statements that HFT is in generaleither good or bad for the market shouldtherefore be viewed with caution.

HFT players and exchange operatorsare at pains to stress that overall HFTperceptibly improves market liquidityand the efficiency of price discovery(McEachern Gibbs 2009). The majorityof investors benefit from reduced bid/askspreads – a common measure of liquid-ity, they say. This statement is backedup by several scientific studies (Gomberet al. 2011). However, there is increas-ing evidence to suggest that, especiallyin very volatile market situations, HFTcould prove problematic and could addi-tionally destabilize the market (Brogaard2010). This must be investigated and, iffound to be true, regulators must step into limit the risks for the financial system.

The flash crash demonstrated that theliquidity generated by HFT market mak-ers, which usually keeps transaction costslow, may suddenly evaporate in difficultmarket phases (NANEX 2010). Unlikeregular “human” market makers, who areobliged to remain in the market even intimes of extremely volatile prices, HFTtraders are generally not bound by suchconstraints. In good times, HFT traderstherefore crowd out normal market mak-ers and often even perform their role bet-ter, to the advantage of all market players.In difficult markets, however, there is arisk that trading flows could collapse withall the attendant problems for the mar-ket as a whole, as HFT players withdraw.To many market participants, the nar-rower bid/ask spreads and higher trad-ing volume generated by HFT thereforeonly represent “sham liquidity”. For thisreason there have been calls from various

quarters to oblige HFT market makers toremain in the market even in times ofhigh volatility, similar to the obligationsimposed on normal market makers (EC2010). In other words, they should startto take some responsibility for the mar-kets which they have, to date, merely usedto their advantage from their superiorposition.

From a regulatory perspective, HFThas proven problematic not only in theserare but dramatic high volatility events,but also in daily trading activities. Whilebid/ask spreads have dropped signifi-cantly in recent years thanks to HFTmarket makers, the average period forwhich such players hold positions hasdropped sharply. According to a study onthe flash crash, most HFT market mak-ers close out their positions after no morethan roughly 10 seconds (Kirilenko etal. 2011). That means that the stabiliz-ing effect in the event of heightened mar-ket volatility exerted by “normal” marketmakers has given way to a “hot potatoeffect”, where falling shares are merelypassed around at lightning speed.

As HFT has become more widespread,the number of buy and sell orders has in-creased dramatically in recent years. Thetactic known as quote stuffing, which isused by several HFT algorithms, is partic-ularly problematic. For reasons of trad-ing strategy, the HF trader places a largenumber of orders per second, only tocancel them again almost immediatelybefore execution. The very high cancel-lation rate this causes leads to a markeddivergence between apparent market liq-uidity and actual trading volume. An in-vestor placing an order in response toa bid or ask is therefore often unableto carry out the transaction at the limitshown. Although the explicit transactioncosts appear low, the implied costs maybe much higher. Apparent market liq-uidity and the size of bid/ask spreadsare therefore not by themselves reli-able indicators of market liquidity andefficiency.

An analysis of 1,172 trading days onthe New York Stock Exchange from 2007-01-01 to 2011-09-14 that was carried outrecently by the research firm NANEXshowed that there were just 35 billion realtransactions for 535 billion quotes. Thequotes-to-trades ratio needed to gener-ate US$ 10,000 in real transaction vol-ume moved from roughly 6–7 at the be-ginning of 2007 to 60–80 in mid-2011.Higher figures indicate a less efficientmarket: more information is required to

100 Business & Information Systems Engineering 2|2012

BISE – DISCUSSION

achieve the same trading volume. Sud-den and dramatic spikes in the numberof quotes are increasingly being observedfor individual US stocks, with individ-ual HFT algorithms generating severaltens of thousands of quotes per secondfor several seconds. Such bursts of activ-ity are frequently accompanied by whatare known as “mini flash crashes”, wheresecurities lose 20%, 40% or even morethan 50% of their value in a space of sec-onds for no fundamental reason, only torecover shortly afterwards. For instance,according to the SEC, the United Stateshas witnessed more than 100 such in-explicable crashes since mid-2010 whichare suspected of being caused by HFTalgorithms.

Sending bids or asks is similar to send-ing spam email: both are virtually free forthe sender, but not for the recipient. For-warding and processing such large vol-umes of data causes a lot of problemsand high costs for exchanges and mar-ket participants. Systems are often over-loaded, which is seen by many observersas one of the causes of the flash crash. Tomake matters worse, certain HFT algo-rithms send some of these quotes only tocause other traders or algorithms to act ina certain way, which they can, in turn, ex-ploit. As a consequence, an ever increas-ing number of institutional investors aretransferring their transactions away fromnormal exchanges to “dark pools”, whereit is usually more difficult to make a profitin HFT.

The above-described criticisms intendto show that HFT is a controversial is-sue, requiring an exact analysis of thedetails. In addition to “market friendly”strategies that regulators regard as posi-tive for the market – for instance, statisti-cal arbitrage – there are also “unfriendly”strategies that are seen as worrying. Oth-ers are basically welcome but when actu-ally applied on the market entail prob-lems and dangers which should be elim-inated. HFT market making is just suchan example.

When considering the ultimate ques-tion of whether there is a correlationbetween HFT and market efficiency, itshould be borne in mind that market ef-ficiency mainly means that the price ofan asset adjusts to fundamental changesin its value rapidly. It is not immediatelyclear how HFT algorithms can contributeto that, as decisions are based only onthe status of the order book in the lastfew seconds or indicators based on tech-nical analysis. A block trade of 10,000

shares between two well-informed largeinvestors represents true price discoveryon the market. By contrast, shifting 100shares back and forth between two HFTalgorithms in innumerous times makesno equivalent contribution to trading ef-ficiency, even if this takes place at im-pressive speed. A market that is mainlydominated by HFT is also a market wheremost orders have lost all connection tofundamental factors. And this correlationbetween price and fundamental value iswhat should, in the main, determine thequality of a market.

Dr. Joachim NagelDr. Rafael Zajonz

Deutsche Bundesbank

References

To: Section 1

Aldridge I (2009) High-frequency trading.A practical guide to algorithmic strategies.Wiley, Hoboken

Brogaard JA (2010) High frequency tradingand volatility. Working paper, Northwest-ern University, Chicago

Dacorogna MM, Gencay R, Müller U, OlsenRB, Pictet OV (2001) An introduction tohigh-frequency finance. Academic Press,San Diego

Durbin M (2010) All about high frequencytrading – the easy way to get started.McGraw Hill, New York

Gomber P, Arndt B, Lutat M, Uhle T (2011)High-frequency trading. Working paper,Goethe-Universität Frankfurt

Gomolka J (2011) Algorithmic Trading: Ana-lyse von computergesteuerten Prozessenim Wertpapierhandel unter Verwendungder Multifaktorenregression. Dissertation,Potsdam, Universitätsverlag Potsdam

Hendershott T, Riordan R (2011) Algorith-mic trading and information. Working pa-per, University of California (Berkeley), Karl-sruhe Institute of Technology

Ye G (2011) High frequency trading models +website. Wiley, Hoboken

Zhang F (2010) High-frequency trading,stock volatility, and price discovery. http://ssrn.com/abstract=1691679. Accessed2012-01-30

To: Section 2

European Commission (2010) EuropeanCommission directorate general internalmarket and services. Public consulta-tion: review of the markets in financialinstruments directive (MiFID). http://ec.europa.eu/internal_market/consultations/docs/2010/mifid/consultation_paper_en.pdf. Accessed 2011-01-19

Gomber P, Arndt B, Lutat M, Uhle T (2011)High frequency trading. Working paper,Universität Frankfurt

Kirilenko A, Kyle A, Samadi M, Tuzun T (2011)The flash crash: the impact of high fre-quency trading on an electronic market.Working paper

To: Section 3

Brogaard JA (2010) High frequency tradingand volatility. Working paper, Northwest-ern University, Chicago

Forbes (2010) Wall Street’s speed war. 2010-09-27

Hendershott T, Jones CM, Menkveld AJ (2011)Does algorithmic trading improve liquid-ity? J Finance 66(1):1–33

Hendershott T, Riordan R (2011) High fre-quency trading and price discovery. Work-ing paper, University of California (Berke-ley), Karlsruhe Institute of Technology

Securities and Exchange Commission (SEC)(2010) Concept release on equity mar-ket structure. Release No. 34-61358, FileNo. S7-02-10

Zhang S, Riordan R (2011) Technology andmarket quality: the case of high frequencytrading. In: Proc ECIS 2011

To: Section 4

Deutsche Börse AG (2011) Several internal pa-pers and statistics. August 2011, Eschborn

Brogaard JA (2010) High frequency tradingand volatility. Working paper, Northwest-ern University, Chicago

AFM/Netherlands Authority for the Finan-cial Markets (2010) High frequency trad-ing – the application of advanced tradingtechnology in the European marketplace

SEC and CFTC (2010) Findings regarding themarket events of May 6, 2010

Zhang S, Riordan R (2011) Technology andmarket quality: the case of high frequencytrading. In: Proc ECIS 2011

To: Section 5

Biais B, Foucault T, Moinas S (2010) Equilib-rium algorithmic trading. Working paper,Toulouse School of Economics

Breuer A, Burghof H-P (2011) An empiri-cal analysis of order dynamics in a high-frequency trading environment. Unpub-lished working paper, Universität Hohen-heim

Cvitanic J, Kirilenko A (2010) High frequen-cy traders and asset prices. Unpublishedworking paper, CFTC

Deutsche Börse (2009) Price list for the utiliza-tion of the trading system of FWB FrankfurtStock Exchange. Deutsche Börse AG

Dombert AR (2011) Hochfrequenzhandel:Nutzen und Risiken neuer Technologienauf den Finanzmärkten. Kreditwesen 18:16–18

Fama E (1970) Efficient capital markets: a re-view of theory and empirical work. J Fi-nance 25(2):383–417

Gsell M (2009) Algorithmic activity on Xetra.J Trading 4(3):74–86

Groth S (2009) Algorithmic trading enginesand liquidity contribution: the blurringof “traditional” definitions. In: Godart C,Gronau N, Sharma S, Canals S (eds) Proc 9thIFIP conference on e-business, e-services,and e-society (I3E 2009)

Hendershott T, Riordan R (2011) High fre-quency trading and price discovery. Work-ing paper, University of California (Berke-ley), Karlsruhe Institute of Technology

Business & Information Systems Engineering 2|2012 101

BISE – DISCUSSION

Maurer K-O, Schäfer C (2011) Analysis of bi-nary trading patterns in Xetra. J Trading6(1):46–60

To: Section 6

McEachern Gibbs C (2009) HFT is mar-ket making evolved, says NYSE exec.http://www.advancedtrading.com/managingthedesk/219501268. Accessed2012-01-30

Gomber P, Arndt B, Lutat M, Uhle T (2011)High frequency trading. Working paper,Universität Frankfurt

Brogaard JA (2010) High frequency tradingand volatility. Working paper, Northwest-ern University, Chicago

NANEX (2010) Analysis of the “flashcrash”. http://www.nanex.net/20100506/FlashCrashAnalysis_CompleteText.html.Accessed 2012-01-30

European Commission (2010) EuropeanCommission directorate general internal

market and services. Public consulta-tion: review of the markets in financialinstruments directive (MiFID). http://ec.europa.eu/internal_market/consultations/docs/2010/mifid/consultation_paper_en.pdf. Accessed 2011-01-19

Kirilenko A, Kyle A, Samadi M, Tuzun T (2011)The flash crash: the impact of high fre-quency trading on an electronic market.Working paper

102 Business & Information Systems Engineering 2|2012