By Theo Hildyard , Software AG
@theohildyard
In the fourth blog of our series outlining the Seven Pillars of Market Surveillance 2.0 we look at how monitoring multiple asset classes can help financial services firms to spot and prevent market abuse and fraud.
Few financial institutions or trading entities now focus on a single asset class, which means that multiple asset classes across geographies must be monitored for abuse. From equities & futures to oil and foreign exchange, rogue algorithms or traders can disrupt markets and bankrupt companies.
Market Surveillance 2.0 is the next generation of market surveillance; acting like a crystal ball to help us see the early warning signs of unwanted human or technological behaviors that signal error or fraud.
In my last blog, Pillar #3: Support for fast Big Data, I outlined how monitoring data now also means keeping tabs on social media, emails, instant messages, news headlines and even audio data from phone calls. For example, the Libor and FX scandals of 2014 may have been prevented by effective monitoring of social chat rooms and instant messaging.
In Pillar #4 we explore the rationale behind needing support for cross-asset class monitoring. While it used to be common to monitor each asset class as a separate entity in so-called silos, it has become clear that many asset classes are highly correlated. In other words, events that impact one asset class can have a knock-on effect to others.
Between 2007 and 2012, for example, the price of oil was highly correlated to the stock market. The S&P 500 marched in virtual lockstep with the spot price of West Texas Intermediate crude oil. Therefore, if something spooked one market the other reacted.
For example, On May 5, 2011, the crude oil market experienced its second-largest daily drop ever when trading algorithms repeatedly triggered sell-stops. The $13 drop in the price of Brent crude knocked the Dow Jones Industrial Average down by 140 points, or 1.1%. It could have been much worse if the oil algos had not been caught and stopped.
Other asset class correlations include bonds and stocks, which tend to move in opposite directions. Usually when stocks go up in value, bonds go down. This is because investors jump into stocks when the economy is booming and sell their bonds, which have a lower return. But even the most tried and true correlations fall apart occasionally, signaling trading errors or perhaps fundamental changes in their relationship.
On October 14, 2014, for example, the U.S. stock market plunged over 450 points before recovering somewhat to close 173 points down. This occurred just after a shocking bond market “flash crash” took 10-year Treasury yields sharply down as liquidity disappeared, and then back up again.
The yield on the benchmark 10-year US government bond, which moves inversely to price, plunged 33 basis points to 1.86 per cent before rising to settle at 2.13 per cent. The FT said: “While that may not seem like much, analysts say the move was seven standard deviations away from its intraday norm – meaning it might be expected to occur once every 1.6 billion years.”
Causes for the bond flash crash are uncertain, but may have been triggered by a rumor in the market that a hedge fund with a very large short bond position was covering, according to Themis Trading. In this bond/stock market crash scenario, it may have been possible to predict some looming anomalies by monitoring Twitter and other social media and apply some algorithmic “shock absorbers” that could protect your portfolio either way the two markets moved.
Today’s markets become more complicated and sophisticated by the day. Flash crashes, trading errors and scandals such as LIBOR, FX and metals fixings mean that financial services firms and regulators must watch all markets at all times across all geographies. Human traders are simply not equipped to monitor and react to multiple asset classes at the same time, trading in lightning speed.
Market Surveillance 2.0 is the answer; taking the fast Big Data from silos of asset classes across time zones and geographies and watching for patterns that signal risk or opportunity, then kicking out real actions to take.
By combining Pillar #1, a convergent threat system, Pillar #2, a combination of historical, real-time and predictive analysis tools, and Pillar #3, support for fast Big Data, with Pillar #4, support for multiple asset classes, we go a long way towards creating our Market Surveillance 2.0 crystal ball.
To find out more, download the full whitepaper here