Why high-frequency trading is a good thingEdward Backes, Head of Market Supervision, Eurex Frankfurt AGThe recent market turbulence and unusually high volatility that came with it have againdirected wider attention to a particular group of trading participants: high-frequency traders –that is, participants who place buy and sell orders in the order book with extreme frequencyand, in the vast majority of cases, do not have any open positions left on their books at the endof the day. The popular perception is that price volatility would be significantly reduced if suchhigh-speed trading did not exist. The question is: Does the media paint an accurate picture andis there any truth in headlines like “Profiteering from Stock-Market Turbulence”, “AutomatedPanic”, and “High-Frequency Trading – Potential Devastator”)?Several research institutions have recently examined high-frequency trading and publishedextensive studies on the topic. Two examples are a study published in Britain by theGovernment Office for Science in September 2011 called “The Future of Computer Tradingin Financial Markets,” and a study by the Goethe University in Frankfurt from April 2011titled “High-Frequency Trading”. Both studies concluded that high-frequency trading isbeing wrongly demonised. In fact, the central message of both studies is clear: Such tradingtechniques actually increase liquidity and improve market quality.As the operator of both cash and derivatives markets, we examined how accurate thesestudies’ findings are – particularly in light of Eurex’s guiding principles of transparency, fairprice determination and orderly trading for all market participants. These principles apply forall market participants, regardless of the trading technology and type of access used.We used the market activity on 25 August 2011 as the basis of our own research. On this day,the performance of the DAX Future (FDAX), which tracks the underlying DAX® index ofleading German blue-chip stocks, made the headlines. What had happened on that day? In thespan of 17 minutes, the FDAX fell by more than four percent and then rose again in thefollowing four minutes by two percent. The rumours circulating in the market afterwardsdrew attention to the potential role of high frequency traders in the contract’s dramatic moves.
Our analysis generated some interesting results. First, the fall in the FDAX was triggered by high volume orders of around 6,000 contracts by institutional clients (“buy side”), which were entered into the trading system as sell positions in small batches with a view to “safeguarding interests”. As shown in Figure 1, a price drop was not triggered by an illiquid market situation. In fact, the high volume orders were processed with small price increments. Average turnover increased in this period to more than 1,700 contracts per minute, far higher than the monthly average of just under 300 contracts per minute. At the peak, as many as 4,700 contracts per minute were being traded – a clear sign of a highly liquid order book. Figure 1: Trading in FDAX futures contracts on 25 August 2011 (one-minute intervals) 14000 5700 12000 5600No. of traded contracts 10000 DAX index points 5500 8000 5400 6000 5300 4000 2000 5200 0 5100 15:45 15:50 15:55 16:00 16:02 16:05 Second, the turnover seen in this 20-minute period was generated by the activities of a wide range of participants. A total of around 200 different participants acted as buyers in this period (in a falling market), including but not limited to high-frequency traders. Around 170 different participants acted on the sell side. The detailed analysis indicates there were up to 122 different participants acting as buyers and 106 different participants acting as sellers per minute (see Figure 2). 2
Figure 2: Number of individual buyers and sellers per minute 4000 4,710 67 Number of active market participants 3000 122 2000 97 108 95 104 87 99 95 97 104 101 86 88 103 110 1000 81 77 67Contracts 69 70 0 55 61 55 74-1000 73 79 67 83 84 73 82 83 88 74 87 80 83 86-2000 83 106-3000 -4,710-4000 15:45 15:50 15:55 16:00 16:02Third, the high market liquidity was in large part provided by the actions of the high-frequency traders, as these participants initially absorbed the major sell positions and thenpassed them on to protect the market. We have been observing this typical trading pattern forquite some time. Moreover, the often assumed acceleration of downward movements throughcomputer-based trading strategies was not observed.This ad-hoc analysis makes it clear: During times of market turbulence, regulated markets likeEurex have consistently made valuable contributions to the fair and orderly readjusting ofinvestment strategies for short, medium and long-term investors thanks to their transparentand reliable market infrastructure. We offer our participants sufficiently large liquidity pools,even in volatile market phases. High-frequency traders also make a valuable contributionhere. They help in processing high volume orders in a way that protects the market by placinga rapid succession of small, non-directional buy and sell orders, thus preventing abrupt pricemovements. It can be demonstrated that participants who employ high-frequency techniques 3
serve as liquidity providers, alongside arbitrage investors and hedgers, i.e. participantscarrying out hedging transactions. Our analysis has shown that high volume sell orders in adifficult market environment found a sufficient number of buyers, allowing them to beprocessed in just a few minutes.Nevertheless, it is essential from the point of view of a market operator to keep track ofchanges in the market and technical innovations, and to react if necessary. With a view to theincreasing market share of automated trading strategies, Eurex built protective mechanismsinto the market structure some time ago. These deal with errors, whether they arise from amistaken entry (“fat finger”), a panic attack by an inexperienced trader, or an erroneousalgorithm. These mechanisms include, among other things, volatility interruptions, real-timerisk management and order limits. For example, volatility interruptions allow us toautomatically stop trading in individual products in response to unusually large jumps in pricetriggered by mistaken entries, stop-order cascades or illiquid market situations. This givesparticipants the opportunity to readjust their market assessment and order management beforetrading restarts. A chain reaction, such as that seen in the U.S. flash crash, would have beenand is impossible at Eurex.In summary: It is unfair and counterproductive solely to blame high-frequency traders forvolatile markets and major price fluctuations. At the same time, high-frequency tradingshould only take place in an appropriate regulatory environment in which benefits and risksare well balanced and sufficient consideration is given to both parameters. Minimumrequirements governing organisation and risk control are particularly important here. Weshould refrain from over-regulating, however. This is damaging and encourages participantsto evade the rules and migrate to less stringently supervised trading venues, thus deprivingregulated stock and derivatives exchanges of important liquidity. 4