Cast Software delivers warning on ‘kill switches’ solution to trading glitches
There is a real danger of increased frequency of outages in trading systems, unless the industry puts in a concerted effort to change the way in which structures underpinning systems are developed, according to Lev Lesokhin, EVP strategy and market development at Cast Software.
The response from Cast, a company that provides analysis of systems architecture, comes in response to suggestions that trading systems implement so-called ‘kill switches’ to automatically disable themselves when they spot a glitch.
Investors, traders and regulators are becoming increasingly aware of systemic risk posed by glitches such as the infamous ‘flash crash’ of May 2010, or the failure of Knight Capital’s systems earlier this year, which cost the company more than $400m over a matter of minutes as computerised trading went haywire.
While many industries face similar IT challenges, it is the issues around trading that differentiate the finanical sector, according to Lesokhin.
There are two areas that need to be understood, he says. The first is the view of functionality: in other words, what does the software do? Lesokhin likens this to looking at a car to see if it has four doors. For trading platforms, questions about unctionality would include asking about the logic driving the platform, and what algorithms are coded into it.
The other view necessary is of the structure. For example, for a car, the questions would be: does it have sufficiently robust ball bearings, or tyres the same size?
The problem is that all platforms are subject to continuous development, which gives rise to concerns about overall stability,as more and more changes are made to the original model. This is happening in the financial sector, as exchanges offer different types of trading.
The effects of complexity are multiplied by the fact that exchanges are competing against each other for business, predominantly on volumes.
This only serves to increase the business value to the exchange operators of their ability to cater to algorithmically driven business. Lesokhin warns that this is setting up the industry for the “perfect storm” hitting those building the trading platforms.
Regulatory issue such as Basel, Dodd-Frank, Ucits and AIFMD are driving IT providers and software developers toward modular based solutions. Being able to bolt on another software module to answer demands in respect of, say, Basel or Solvency rules indicates flexibility is required in the design of systems.
However, modularity goes against the objective of those developing trading platforms, which is to maximise performance. Modules have to be able to ‘talk’ to each other, which takes more time. On the other hand, it is harder to implement changes to single, all-encompassing systems.
The battle can be framed in terms of flexibility versus speed, but there is also continued pressure to deliver faster time to market. This creates another set of risks around the implementation of solutions, because it shortens time for quality assurance (QA) testing before going live.
The financial industry is able to attract the top programmers, but the complexities of the platforms being developed have reached a level whereby they are outstripping even these technology superstars. And cost pressures on the financial industry make it tempting to consider amending or limiting QA. Therefore, it is time that regulators recognised the risks developing and put control measures in place, Lesokhin suggests.
This could take the form of a regulatory baseline, to stop competition between exchange platforms without adherence to minimum risk tolerance levels and QA.
“It is not good for platforms to compete without paying heed to structural risks in their systems,” he says.
Picking up the car analogy again, Lesokhin likens it to the US Federal Highways system, where it requires surveying to identify weaknesses. Similarly, those designing and implementing systems should be required to identify any weaknesses in what has become key infrastructure. He believes the industry will eventually recognise this need, but until then the market will continue to see glitches.
The so-called kill switches being proposed by some regulators, to ensure that trading can be halted if it looks as if a glitch is happening, are “a band-aid solution”, Lesokhin says. Instead, he expects increased frequency of serious trading glitches and, as a result, drastic rules being implemented against quantitative trading and other fundamental changes in areas such as high-frequency trading, if the industry does not improve its record.
Regulators are also becoming aware of the danger that events are less likely to occur in isolation in future. Lesokhin cites the example of Bats, which along with other multilateral trading facilities and Nasdaq were ensnared in a trading glitch that sent Kraft Foods Group shares up about 30% in one minute in early October. This served as a warning of the interconnecting of systems, creating domino-effect scenarios.