skip to log on skip to main content
Article related to:

Technology

Algorithms and the digital you don't see

Former Director Business Projects Global Markets, ANZ

2016-03-29 14:52

When we think 'digital', we often think of beautiful apps and websites containing cutting-edge functionality, delivered via simple user interfaces striving to create a flawless experience and, in doing so, encouraging customers to engage.

" Perhaps one day a company's dedicated treasury algorithms will identify an impending currency liability by reviewing the balance sheet and cash flow forecasts."
Todd Tobias, Director of Business Projects at ANZ Global Markets

But there's a hidden side to digital: where companies plug their systems straight into each other. And where the beauty is not in the design of the user interfaces. Instead, it's in the elegance of the programs and calculations carrying out increasingly complicated actions on their owners' behalf. At the core of these activities are algorithms.

This type of deep integration is happening across almost every industry. Within financial markets, improving technologies are expediting the global distribution of individualised products and pricing.

Literally hundreds of price updates every second flow directly into exchanges and electronic communication networks, where banks automatically compete for business - and where orders are placed, priced and filled all within the blink of an eye.

DRIVING TOWARDS DRIVERLESS

Algorithms (a set of calculation rules to be followed by a computer) are constantly being improved and strung together in ever more complex chains, handling more sophisticated possibilities in order to operate, and prevail, within real-world conditions.

{CF_IMAGE}

Source: Yauhen_D / Shutterstock.com

This evolution is evident within cars: what started simply as a rear-view camera, supplementing the perennial rear-vision mirror, added electronic parking guidance and is now available as completely hands-free automatic parking.

A fully autonomous driving capability is still a few years off, though reading the Google Self-Driving Project's monthly reports gives the feeling it is only around the next corner.

But just as algorithms within cars are developing, so too are financial algorithms, from the most basic 'show me my account balance' and 'pay this supplier' activities within Internet banking, through orders to buy or sell a stock when a particular price is hit, right up to complex short-term automated trading algorithms which self-adjust according to market activity.

Algorithms are even able to digest human language news stories and social media and instantly act on events (albeit sometimes with less-than-perfect results), and automated financial advice is beginning to be taken very seriously as well.

Maybe the 'driverless' equivalent in finance will be fully-automated financial management algorithms designed to achieve broad financial objectives, instantly adjusting to changes in the market and other circumstances along the way. These could be tailored to work for individuals, businesses and corporations, even at the level of central banks.

However, just as questions of culpability are being heavily contemplated in the anticipated occurrence of the driverless car accidents to come, so too are questions of liability and control arising as human hands are increasingly removed from every single financial decision and action - and regulators are already developing specific laws around the use of trading algorithms.

Despite regulatory controls however, things will inevitably go wrong: a car crash in one situation, a flash crash in another and virtually any system or algorithm can theoretically be hacked and retasked to nefarious purposes. 

ALGORITHM?

The word Algorithm comes from the name of a prominent Persian mathematician and polymath from the eighth and ninth centuries AD, Muhammad ibn Musa al-Khwarizmi('algoritmi' in Latin); a towering figure in early mathematics who played a major role in the introduction of both decimals and algebra to the Western world.

Humans will therefore be firmly at the wheel of both cars and finances for a while yet.

Although interestingly, an alternative view is the relatively error-prone and therefore unsafe drivers we humans have shown ourselves to be could perhaps one day be banned from being behind the wheel altogether in the face of presumably much safer automated drivers.

So maybe some riskier and/or more delicate financial activities could one day be placed by regulators deliberately beyond the direct control of humans, too.

NO LONGER NICHE

Since sprouting from its earliest beginnings in equities on the New York Stock Exchange in the 1970s and 1980s, until recently algorithmic trading in the global Foreign Exchange market was still a niche activity.

Today it is increasingly used by banks, asset managers and hedge funds and due to its cost and precision benefits is expected to heavily penetrate other areas of the marketplace such as corporate treasuries.

This year for the first time, algorithmic electronic (system-to-system) FX trading volumes could actually surpass non-algorithmic electronic (person-to-system) FX trading volumes.

According to Luke Marriott, Global Head of Wholesale Foreign Exchange at ANZ, the reach of ANZ's automated FX trading channels “has extended from our systems directly into customer and counterparty systems and we now transact about $A5 billion of FX flow per day in this way – and in the course of doing so, we capture and process about 300 million price events".

{CF_IMAGE}

Perhaps one day a company's dedicated treasury algorithms will identify an impending currency liability by reviewing the balance sheet and cash flow forecasts, and using inputs from a range of economic modelling algorithms, determine the most advantageous method of funding the liability.

Communicating directly with the company's banking partners, treasury algorithms will then trigger the optimum foreign exchange execution algorithm in order to fulfil the particular currency requirements – at the right time and the lowest cost.

Even before it even instigates a real transaction, a treasury algorithm might utilise the immense power of inexpensive cloud computing and spend a few short minutes simulating the whole process a few thousand times in advance to refine the funding strategy and tweak its execution parameters.

ET TU, ROBOTS?

Humans won't be relegated to the sidelines just yet. We will form and manage the relationships between the organisations, determine the strategic objectives and develop the algorithms that deliver them (well, until such time as the algorithms start developing themselves), and will conduct essential governance and control functions too.

In fact, the Automation Paradox says as human involvement reduces, the remaining human roles become increasingly important, principally in order to catch errors and unintended consequences before they replicate and amplify; anybody familiar with the plot of The Terminator will understand.

Ultimately however, nobody quite knows the extent to which algorithmic automation will take over in finance or the forms it will ultimately assume, although there are two things of which we can be certain: the world of finance is not going to become any less automated than it is today, and the next few years will bring an exciting and twisting network of fast lanes, tight turns and dead ends.

Todd Tobias Director of Business Projects at ANZ Global Markets.

The views and opinions expressed in this communication are those of the author and may not necessarily state or reflect those of ANZ.

anzcomau:Bluenotes/technology-innovation,anzcomau:Bluenotes/technology-innovation/culture,anzcomau:Bluenotes/technology-innovation/digital,anzcomau:Bluenotes/technology-innovation/disruption,anzcomau:Bluenotes/technology-innovation/entrepreneurs,anzcomau:Bluenotes/technology-innovation/innovation,anzcomau:Bluenotes/technology-innovation/tech
Algorithms and the digital you don't see
Todd Tobias
Former Director Business Projects Global Markets, ANZ
2016-03-29
/content/dam/anzcomau/bluenotes/images/articles/2016/March/driverless-car-thumb.jpg

EDITOR'S PICKS

Top