A significant issue with High Frequency Trading (HFT) has always been defining the activity and related behavior during times of crisis.
“The issue is HFT impact on market stability,” noted Dr. Randolf Roth, Head of Market Structure for the Eurex Exchange.
With the May 6, 2010 Flash Crash in mind, Eurex exchange officials detailed their proprietary “Volatility Interruption” algorithm. The algorithm is essentially a trailing look-back trigger that provides markets a cooling off period when abnormal volatility is detected.
“This is different from a traditional exchange circuit breaker,” said Vassilis Vergotis, Executive Vice President, Head of Eurex / Americas. Exchanges traditionally stop trading simply based on a given percent price move in a giving market. For instance, if the price of corn were to trigger a circuit breaker in the S&P 500, for instance, trading might stop once a 20% down move in price occurred. Such “lock limit” moves are imposed on a daily basis, on an exchange by exchange basis, and did not prevent the Flash Crash due in part to the speed in which a crash can gain momentum. In an electronic trading world, the use of high latency “electronic eye” technology is utilized by HFT and electronic market makers to monitor and predict liquidity disparity and volatility. The key is the speed with which a market flash crash can gain momentum quickly and on small volatility spikes.
Mathematical Logic Drives Protection
Noteworthy is the sophistication of the Eurex algorithm. Moving past simple daily price limits, the algorithm has a volatility detection feature that operates over a shorter time period. If abnormal or potentially damaging crash volatility is detected, a systematic decision process ensues that stops market trading until calm can be restored. This is important in avoiding a volatility flash crash because it could diminish the potential for cascading stop orders to be executed.
The exchange says the algorithm has been used on average under ten times per year in the derivatives markets, while its use at the Deutsche Boerse stock exchange might occur several times per day.
Looking at the distribution of liquidity, the algorithm provides traders a “cooling off period” that allows calmer heads to prevail. “We’re trying to avoid the issue of cascading stop orders,” Mr. Vergotis said. Cascading stop orders occur when a market experiences volatile price moves in a short period of time, triggering automated sell or buy orders. Such market environments can trigger cascading stop orders which, in extreme circumstances, might trigger “Flash Crash” type circumstances.
Discussing the sophistication of the algorithm, Eurex officials addressed the mathematical model has relative overrides while not disclosing the specific math. Interestingly, the interruption algorithm was a variable defined by how the trading day started. For instance, if trading on the day started without volatility but then experienced a sudden pick up in volatility, the algorithm would shut down trading quicker than if the trading day started with volatility and then moved to higher volatility. Further, if the algorithm detected abnormal distribution of bids and offers, this trigger could shut down trading for a variable length depending on the significance of the volatility. The average length of a market shutdown might be 2-3 minutes, upon which an auction process precedes stabilization of distribution of market bids and offers.
While no system is perfect or guaranteed, particularly under flash crash circumstances, risk managers are advised to stress test mathematical logic during periods of market crisis.
DISCLOSURE: These are the opinions of the author and may not have considered all risk factors. Nothing on this web site should be construed as an individual recommendation, talk to your independent advisor. The author and Opalesque may have relationships with those people they cover in the publication. Mr. Melin provides a full disclosure of his business relationships to regulators and certain eligible participants who engage him in consulting projects. Managed futures investing involves risk and there are no guarantees of safety or future performance being implied. Managed futures can be a risky investment. This web site and its content is subject to the terms of the web site. Risk Disclosure and terms of web site are available here: http://www.uncorrelatedinvestments.com/templates/Disclaimer.html Performance information received on this site is provided by third parties and deemed reliable but there is no guarantee relative to same. Performance reporting sources and quality assurance techniques may include, but are not limited to: disclosure document, CTA self reporting, brokerage firm reporting, consultant reporting, spot checking other reporting databases; nonetheless no guarantee of accuracy or implication performance verification or auditing is being made by the publishers. The CTA Database is a project separately managed from www.uncorrelatedinvestments.com.