Will Your Next Fund Manager Be A Super-Computer?
15 September 2017 | World Views | AJ Cilliers

Much has been made in recent times of the March of the Machines.  The narrative suggests that human beings are facing a triple threat, made up of big data, super-fast computers and smart algorithms.  This is expected to change the working worlds of many professionals, from medical diagnosticians to fund managers.  It is the second group that concerns us here: if fund managers with their massive resources can’t beat the machines, what chance do individual investors have?

To understand the issue, we need to appreciate that humans are generating massive amounts of new data on a daily basis.  What is more, the speed at which we are generating data is increasing, so that today we create 90% more data daily than we did two years ago.  This growth curve is likely to flatten out at some point, but there is no evidence for this at present. 

The implications of this data explosion are undeniable; as Linda Kool of the Dutch Rathenau Institute points out: "The ’data-driven society’ is on its way. The explosive growth in the quantity of digital data has provided a resource for innovation, leading to new economic and social opportunities."

Superfast computers

Of course, all of this data (for example, all of the share price movements on all of the world’s major stock markets since their inception) is only useful if there is the time and capacity to analyse it.  Until quite recently this was not the case; computers weren’t fast enough, so that the ability to analyse massive databases was of little practical concern. 

However, in line with Moore’s Law, computing power has doubled roughly every 18 months over the past 34 years, making superfast computers a reality today.  What used to take weeks or even months to analyse can now be performed in split seconds.  And technological advances promise further huge strides forward.  Quantum computing is now on our doorstep, and in recent months computing seemed to leap even further ahead, to enter the realm of science fiction.

Professor Ross King, from the University of Manchester’s School of Computing, made an astonishing announcement in March of this year.  He said the following: "Imagine a computer is searching a maze and comes to a choice point, one path leading left, the other right.  Electronic computers need to choose which path to follow first, but our new computer doesn’t need to choose, for it can replicate itself and follow both paths at the same time, thus finding the answer faster."

"This ’magical’ property is possible because the computer’s processors are made of DNA rather than silicon chips.  All electronic computers have a fixed number of chips.  Our computer’s ability to grow as it computes makes it faster than any other form of computer, and enables the solution of many computational problems previously considered impossible.” 

So quantum computers will be exponentially faster than the current electronic computers, and DNA computers will be faster still.  Even if data does keep growing at scary rates, it seems that computing speed will keep up.

Smart algorithms

This leaves us with the third development, that of smart algorithms.  As databases grew in size and computing speed improved, a new generation of geeks entered the field.  These were the young Maths wizards, increasingly with PhDs in the discipline, responsible for developing the tools that would allow super-fast computers to analyse huge databases.

Algorithms have been around for a long time, but smart algorithms are a more recent development.  We all know that computers have to be programmed to carry out certain tasks.  This involves writing a programme, which is a series of step-by-step instructions telling the computer what to do.  Now when you tell the computer what to do, you also need to tell it how to do it.  The how part is where the algorithm comes in. 

In the past, algorithms were also a set of step-by-step instructions, but smart algorithms are different.  Smart algorithms are designed in such a way that computers can learn on their own (i.e. these algorithms facilitate machine learning).  To understand the implications of smart algorithms in the world of share trading, we’ll now look at a specific example.

A US-based developer of trading algorithms, Pragma, has developed an algorithm called Ghost.  Ghost can automatically increase or decrease the pace of trade in a particular share, based on prevailing conditions in the market.  These conditions could include the volume of shares being traded, the volatility of the market, and movements in the share price. 

The end result is that Ghost is able to ensure a buyer the best price in the market at any given time, and also to avoid the trade being anticipated by others.  This has led many large buy-side investors, such as pension funds and unit trust funds, to use such algorithms.

The current strength of super-fast computers is that they can be used to analyse a vast pool of diverse (’big’) data to find correlations, or relationships, between different variables.  So, for example, a super-fast computer can be programmed to test the statistical relationships between changes in the price of a share, and changes in a whole range of other variables. 

If unexpected relationships are discovered, for example between share price and the pace of trades in that share, the volume of the shares being traded, and the volatility of the share market as a whole (as per the example above), then a smart algorithm can be developed to take advantage of these relationships.

Are fund managers a dying breed?

So far, we can see that from a trading point of view it may be very difficult to beat the machines. Using smart algorithms, automated trading systems can monitor a variety of different market factors, then pick the exact moment to execute a trade and get the best price. However, what the machines cannot yet do is pick winning shares - shares that will perform better than a selected market index over a period of time.

When it comes to beating the market, it seems that some humans still have the upper hand.  Computers can run passive funds that will match a market index, but some humans can do even better.  For example, in a recent evaluation of 426 active fund managers in the UK, 100 were found to have outperformed the market over their entire careers to date, in more years than not. 

The average monthly outperformance of this group has been 0.38% per month, or 4.56% per year.  If we focus in particular on the top 10 of the 100, we find that they have all had long careers, with the lowest being 8 years, the longest being 26 years, and the average career spanning 18.1 years.  The best of these has averaged a monthly outperformance of 0.97% per month for 16 years, while the 10th best performer managed 0.42% per month over a 24 year period.

Of course, the outperformers make up less than 24% of all active managers, prompting successful fund manager Mark Asquith to say: "It’s a sad fact of investment life that that some 75% of active fund managers have under-performed a passive index over the past five years.  I call these underperformers the slaughtered three-quarters.’  They fail to perform because too many of them are not active enough.  They’re not stock pickers."

This is an interesting point, and is supported by research.  This shows that unsuccessful fund managers often select a portfolio which will match their sector or market index.  In other words, they try to construct the equivalent of an index fund.  They then go ’bigger’ on a small number of shares, chosen on the basis that these shares will considerably outperform their chosen index, and thereby lead to outperformance for their fund as a whole.  These traders don’t actively pick too many shares, and they limit their trading to keep costs low.  And for most of them, the strategy fails.


At present, certain active fund managers are beating the machines and achieving outperformance.  What we haven’t examined yet is exactly how they do it.  Next month we’ll address this issue, and also look at the cognitive style most likely to produce favourable results, for stock-picking fund managers and individual investors alike.



AJ Cillers

AJ is an academic and a freelance financial journalist who has written for Sharenet for some 15 years. He spent 25 years as an accountant and financial manager in various South African companies before moving into academia. He has a broad range of interests, including all aspects of business and stock market investing. Apart from a bachelor’s degree in Accounting, AJ holds a Master’s degree in Financial Management. He is also a Fellow of the Chartered Institute of Management Accountants.

The information contained in this article is for informational purposes only and must not be regarded as a prospectus for any security, financial product or transaction. It is neither to be construed as financial advice nor to be regarded as a definitive analysis of any financial issue. Investors should consider this research/article as only a single factor in making their investment decision. We recommend you consult a financial planner/advisor to take into account your particular investment objectives, financial situation and individual needs. The views and opinions (where expressed) in this article are those of the author and do not necessarily reflect the official policy or position of Sharenet.

Share This Story

Market Statistics are calculated by Sharenet and are therefore not the official JSE Market Statistics. The calculation/derivation may include underlying JSE data.
© 2018 SHARENET (PTY) Ltd, Cape Town, South Africa