Credit suisse declares high-frequency trading (hft) to be “harmful”

Credit suisse declares high-frequency trading (hft) to be 'harmful'schadlich'

Apparently, it is not the HFT trades that are the problem, but the orders that are not filled. Source: nanex

The widely used HFT strategies “quote stuffing”, “layering” and “momentum ignition” led to market manipulation, which is already prohibited. Further regulation would therefore be superfluous

Although CS analysts insist that not all HFT strategies are harmful, two recent reports high frequency trading – measurement, detection and response and high frequency trading – the good, the bad, and the regulation nevertheless paint a picture that confirms everything that has been asserted in these pages for years as well.

To do this, they analyzed all the stocks traded by CS – the swiss universal bank is also a leading global brokerage house thanks to its subsidiary first boston – to see whether and how often they are targeted by HFT and what influence this has on the prices and the prices quoted. This was not done for analytical reasons alone, but primarily because CS wants to protect itself and its clients from being taken to the cleaners by HFT, or at least to demonstrate its ability to do so. Or at least to demonstrate that it is capable of doing so. After all, it has long been suspected in stock market circles that the consistently declining interest of real investors in the stock markets can be explained not only by the two severe global collapses since 2000, but also by the fact that ordinary stock market investors are now faced with technically far superior opponents who ensure unfair market conditions.

HFT’s share in european equity trading. Source: credit suisse AES analysis

The study was prepared by CS’s advanced execution services (AES) department, which was established in 2001 as the first such department on wall street for program trading (algorithmic trading). AES prides itself on providing its exclusively professional clients with the latest technology and programs, as well as electronic access to 75 exchanges and liquidity pools worldwide, with the stated goal of achieving the best possible prices through program trading, even for large volumes. But even though the report obviously serves to demonstrate how well AES can hold its own against HFT, it also allowed CS to be the first relevant market participant to provide a detailed insight into what is really happening in the equity markets based on its own business.

On the one end of the spectrum, market making and statistical arbitrage strategies profit by reacting to – and correcting – short term mispricing, thereby improving market quality. In contrast, other strategies such as quote stuffing, latency arbitrage and momentum ignition seek to create short term mispricing and subsequently profit from it.


Thus, AES seems to be in a permanent race with the HFT, for which it permanently updates signal analysis techniques like pattern recognition or “real time burst detection” to detect and analyze HFT activity in the securities traded by CS. If it now proves that many of the HFT strategies are in fact “harmful” this is grist to the mill of those calling for stricter regulation, which of course could not be in the interest of CS.

As a reminder: from an economic point of view, it is clear that the main source of profit for HFT can only be the lower prices that other, less fast market participants get or have to pay due to HFT activities. Have to pay. HFT, on the other hand, argues that they have been “provide liquidity” which, according to CS, is indeed the case for some of the HFT strategies, which are referred to as market making strategies, such as when hfts uncover and eliminate price anomalies, as is done in classic arbitrage trades.

Since the emergence of HFT, according to AES, there has been a general narrowing of spreads (difference between bid and ask prices), which could also be related to the fact that the normal stock exchanges have also increased enormously and additional mfts have emerged, which execute orders across as many trading venues as possible. According to a swedish study, benign strategies are said to account for around two-thirds of HFT activity on the stockholm stock exchange, but the remainder is clearly “harmful” strategies, three of which the analysts analyze in detail and largely agree with the financial information analyst nanex, who until now has been almost solely responsible for qualified information about the HFT sector.

“quote stuffing”, “layering”, “momentum ignition”

On nanex also goes the name “quote stuffing” back, which was allowed to be probably the most important of the bosartigen HFT strategies. In this process, the HFT sends rough volumes of orders to the stock exchanges, but these are cancelled again so quickly that although a rough number of new “best bid”- or “best ask”-orders are created, but cannot be executed. Apparently, this affects almost all continuously traded shares.

Accordingly, the 600 stocks listed in the eurostoxx 600 index in the 3. In the first quarter of 2012, quota stuffing affected users 18.6 times a day on average, with 54.6% of attacks being over after two seconds. On the other hand, 27.9 percent lasted longer than one minute, with volatility and spreads temporarily rising significantly after the attacks.

HFT attacks occur mainly in the early morning and around the release of important data, with the chance of further attacks increasing with each attack, usually occurring within the next five minutes. This could, for example, disrupt the systems of other trading participants or cause other traders to make incorrect (d.H. Favorable for the HFT) orders. So-called black pools, where larger blocks of shares are not traded through normal stock exchanges, but through unregulated pools of large banks such as.B. Also be traded CS. This is because trading often takes place at the mid-price between current bid and ask quotes, so that buyers and sellers should actually share the spread, which is one of the main arguments for trading via black pools. The hfts now try to produce a false mid-price, so that they can collect the spread alone.

At “layering” the HFT enters a rough number of sell (or buy) orders to create the impression of. Buy orders) in order to create the impression of rough selling prere. If the price starts to move, the HFT buys cheaply and then immediately cancels all orders, which can bring high profits if the HFT is actually fast enough not to be filled. The trader on the other hand does not receive the price he saw on the screen, but a worse one, which leads to complaints from affected customers at CS, who have to realize that immediately after their trade the previously full order book is suddenly empty”order book fade”), which may also extend to other trading places.

With “momentum ignition” hfts try to entice as many other market participants as possible to trade quickly in order to drive the price in a certain direction. The goal is then to sell a previously existing position at a profit or to profit from it when the price change it has caused recedes again. The stoxx 600 stocks were affected on average 1.6 times a day during the period, with HFT-induced price changes averaging 38 basis points (100 bp = 1 percentage point). In five percent of the attacks, however, the jump reached more than 75 bp, and in just under one percent it even exceeded 500 basis points, which already promises considerable returns, especially when one considers that such actions are over after 1.5 minutes on average and the average bid-ask spread for stoxx600 stocks is eight basis points.

To counter such strategies, CS has meanwhile developed a number of tools that are intended to ward off such attacks. However, if the aforementioned malicious strategies were indeed market manipulations – which the authors leave no doubt about – they were already prohibited by law. However, with a few exceptions, such cases have not yet been prosecuted, which is probably due to the fact that the investors who have been harmed either do not notice this or are unable to prove it.

According to the authors, this could be remedied, for example, by setting a quota for the minimum number of orders entered that must be executed, as is already prescribed with moderate success in france and germany and by a number of stock exchanges. In order to combat the already illegal practices, this had to be seriously attempted according to CS blob. This is better than waiting a long time for new legislation or even making snap decisions, which analysts fear will create excessive bureaucracy and eliminate the positive effects of HFT.

However, in order to facilitate the prosecution of these practices, the authors seem to expect a corresponding practice in the upcoming new EU financial market regulations mifid 2, which, according to the relevant ECON committee of the EU parliament, should even impose a minimum duration of half a second for each order. However, the authors seem to think that it is unlikely that this idea will be implemented, because if it were to happen, HFT would profit from the outdated quotations, and soon no one would be interested in placing orders in the system at all. This was expected to significantly reduce market liquidity, with the authors citing the chief market strategist at deutsche borse and the UK government foresight project. Presumably, however, the defense potential of HFT, which had been built up laboriously and expensively, would then also become obsolete, which is why credit suiss can ultimately hardly be regarded as a serious opponent of HFT.