Tech Beat: Rethinking exchange reliability

One of the hazards of trading is the frequency at which venues break down.

In the last month alone there have been minor glitches at Nasdaq, Singapore exchange SGX, Euronext, Bloomberg and several times at the Chicago Board Options Exchange. There are more.

Add in the high profile problems that befell the listing of Facebook, the troubles at Knight Capital, the alphabetic trading glitch at Goldman Sachs, the listing of BATS Global Markets and the three-hour Nasdaq outage last summer, and a depressing picture of reliability quickly emerges.

Two London-based entrepreneurs have had enough of this sorry pattern and set up a business that offers a radically different approach to developing the tools for trading.

Grant Passmore and Denis Ignatovich, an academic mathematician and a Wall Street trader respectively, have founded a company called Aesthetic Integration that aims to reshape how the industry builds its own technology.

"The software has become so complex, no one is sure how it all works," says Mr Ignatovich of the trading industry.

Aesthetic Integration explores the use in financial markets of a fast-emerging trend known as formal verification. Simply put, formal verification uses advanced mathematics to ensure that the algorithms used in high tech engineering designs conform to specified requirements. Such a state of the art approach adopted in trading technology, its advocates say, could eliminate glitches before they reach the market.

And unlike some other tech trends, formal verification builds on serious real-world experience, which began with a debacle.

Twenty years ago, Intel released a Pentium chip that contained a serious bug, forcing the chipmaker into a recall and redesign that cost it a hefty $475m and public embarrassment.

As it sought to resolve the issue, Intel hit upon academic mathematical work to run formal checks by algorithms. Computers now verify Intel's designs at every stage of the process before it makes a chip. The process has uncovered bugs of a range of sizes, from the subtle to the very serious.

The aerospace industry uses formal verification to check the safety of complex software systems underlying both air traffic management and autopilots.

Microsoft has used it after the heavy criticism of its Vista operating system and requires that critical codes destined to drive pieces of hardware must pass its own verification tool.

Apple is adopting a similar approach and is hiring engineers for key positions. Now the technique is being used to test automated transport systems such as railways and self-driving cars.

So why has formal verification not been used in markets before?

One factor, says Mr Passmore, has been its complexity. "Historically it has cost a lot of time and money." Another, Mr Ignatovich notes, is that exchanges and brokers have not thought about themselves as manufacturing companies. "Complexity has tweaked [their systems] so much," he says.

Furthermore, in recent months there have been mathematical breakthroughs - to which Mr Passmore contributed - in a field known as satisfiable modulo theories.

It has meant mathematicians can apply formal verification to large-scale software systems that consume lots of unstructured data - one of the main characteristics of a financial market trading venue.

This has allowed Mr Passmore and Mr Ignatevich to develop a specialist tool, called Imandra, that is tailored to trading systems. The duo have used it to compile just 300 lines of computer code that can test compliance with the SIX Swiss Exchange's 200 pages of guidance on functions and order types.

"We can write out the language for an exchange into the language of the tool," says Mr Passmore. The company is in the early stages of talks with exchanges as they have to publish such user guides but in time it could be extended to brokers and investors.

Nevertheless, the results of the formal verification technique in general are not perfect.

Intel chips can still have flaws, as its recall of the Sandy Bridge chip set in 2011 showed, but they have been minimised. New developments have yet to be tested.

"These breakthroughs are huge but there's still so much to be done," acknowledges Mr Passmore.

One cannot predict how successful Aesthetic Integration will be in bringing the approach to financial markets. Nevertheless, one cannot ignore the spread of the approach and with that list of high profile trading technology failures getting longer by the week, it is surely only a matter of time before the financial industry explores it.

© The Financial Times Limited 2014. All rights reserved.
FT and Financial Times are trademarks of the Financial Times Ltd.
Not to be redistributed, copied or modified in any way.
Euro2day.gr is solely responsible for providing this translation and the Financial Times Limited does not accept any liability for the accuracy or quality of the translation