It sounds like a plot line plucked from the pages of an Arthur C Clarke
science fiction novel: take a computer that makes its own decisions, independent
of human beings, and let it try to beat the fallible traders in the financial
For those at the cutting edge of financial markets, the use of highly
sophisticated computer programmes, known as algorithms, is far from fantastical:
it has been a key driver for the success of some of the world’s largest hedge
Over the past four years, for example, the chief executive of Olive Tree
Capital, James Casper, has been adapting a programme designed for the Israelis
to track enemy fighter planes. Now he’s using it as the secret weapon of a newly
launched foreign exchange hedge fund.
One of the most profitable hedge funds is Renaissance Technologies. It has
around $26bn under management and has yielded returns exceeding 30% for more
than a decade. Clouded in secrecy, it is almost entirely staffed by scientists
that use computers to make money.
While Olive Tree Capital and Renaissance Technologies invest in commodities
and foreign exchange, the use of algorithms to implement quantitative strategies
in equities has also grown.
John Godden, the chief executive of the IGS Group, a hedge fund advisory
firm, explains: “The funds invested in equity quant strategies have grown
because they are one of the better ways of diversifying portfolio risk.”
The growth of the use of algorithms to buy and sell shares poses an interesting
question for financial executives: if the black boxes are making the trading
decisions, why bother spending time and money on investor relations?
Although the idea of computers, rather than humans, making decisions about
when to buy and sell could make any executive paranoid about the machines taking
over the world, there is no need to panic. The increasing use of algorithms can
help a company’s share price.
“Us quants can help finance directors. We make their markets more liquid
through the sheer number of the buy and sell orders that we make at more
consistent prices. Everyone loves a liquid market; it feels like a more even
playing field,” says Casper.
Quant and leap
Although improving computational power has been a key to the growth of
algorithms implementing quant strategies, an equally important tool has been the
improvement in communications technology.
“For the algorithm to work, it needs to be able to get hold of the data from
a supplier. Five years ago you could not get the data in the right format. Now,
not only can you get the data in the right architecture so it can be processed,
you can also send information so the algorithm can execute its buy and sell
orders,” says Casper
Download our Whitepapers
The systems and infrastructure have developed to the stage where they are
powerful enough to allow algorithms to trade up to several hundred times a day.
These algorithms implementing quant strategies can also provide important
liquidity when the share price of a company is falling, says Godden. “When you
really have a run on the stock, these algorithms are likely to have picked up
the trend earlier on than the fundamental investor and would start buying back
the stock before the discretionary funds,” he says.
This ability to buy and sell in a matter of minutes does increase the
volatility of the stock, although it does not have an impact on the actual share
price, says Godden.
In theory, if a number of different trends gang up, this could cause a run on
a share price, but the likelihood of this happening has diminished as algorithms
have had to become increasingly sophisticated.
Statistical arbitrage is a good example of how fund managers have had to
evolve their investment strategies quickly. It is the most popular quant
strategy in the hedge fund universe. Jim Simmons, president of Renaissance T
echnologies, is a champion of statistical arbitrage.
From 1995 to 2000, statistical arbitrage was the darling of the hedge fund
industry, making excellent returns. But it has become more difficult to make
money from it as the growth in the number of players in the market made the
trends dry up more rapidly.
Back in the good old days, statistical arbitrage was very simple: it looked
at the correlation of two share prices – Tesco and Sainsbury’s, for example.
When one share price moved out of whack, then the trader would buy or sell the
shares to make a profit when normality resumed.
For statistical arbitrage to be successful today, hedge fund managers have to
look at the correlation between much more than two share prices. Now 1,000 share
prices are compared. Or a large number of variables from many types of asset
classes: for example, the spot oil price, the dollar-yen exchange rate and
Indian share futures.
Brad Bailey, a senior analyst at the consulting firm Aite Group, explains:
“What is fascinating is how quickly things are evolving and how quickly quant
strategies change. Whereas certain trading opportunities might have, in the
past, worked for a few months, they now only work for a few days. It requires a
tremendous intelligence and agility to constantly jiggle your algorithms for
As the strategies have become increasingly sophisticated and covered an
increasing number of asset classes, so the probability of a large number of
strategies causing a run on a stock has diminished.
But the use of algorithms to implement increasingly sophisticated hedge fund
strategies is only half of the story: computers are also rapidly changing the
way shares are bought and sold.
Algorithmic trading is changing the look and feel of trading floors around
the world. In Andrew Marr’s BBC programme, A History of Modern
Britain, there was footage from a trading floor on Black Wednesday in 1
992, when Britain exited the exchange rate mechanism. As the day’s events
unfolded, traders sweated and looked dumbstruck, then stirred themselves into
action, scurrying around the floor and shouting into phones.
If a similar crisis were to unfurl today, there would be a much more muted
response: many traders have been replaced by beige boxes housing powerful
computers that silently issue buy and sell orders.
The London Stock Exchange estimates that around 40% of its electronic order
book is now carried out by algorithmic trading programmes. A recent study by the
IBM Institute of Business Value predicted that for every 40 traders in a
financial asset class only four would remain by 2015.
When it comes to trading, computers have the edge over human beings. Take
reaction time as an example. A human trader could probably react to the change
in share price in a few hundred milliseconds. A computer can do that at least
ten times faster.
A computer is also capable of monitoring thousands of share prices and
carrying out multiple trades, all at super-human speed.
Traders have good reason to believe they will be replaced by the machines,
but algorithmic trading poses less of a threat to the FD; in fact, it can even
help them to keep dealing costs under control, particularly with large share
Algorithmic trading allows a company to ensure that its share buy-back is
traded at the best price. Toby Bayliss, head of algorithmic sales in Europe for
Citi, says, “If you use an algorithm to carry out a trade, it will determine the
best way to carry out the order, including how to slice it as well as when and
where to send each of these slices. As it deals with real-time data from the
market, the algorithm provides statistically more consistent results than a
One of the biggest problems executing a large share buy-back has been keeping
knowledge of the deal’s execution from the marketplace. “If you gave it to the
cash trader, the potential for information leaking into the market is far
greater, increasing the possibility for the market to move against you,” says
If the order is executed by an algorithm, there is less interaction and it
also takes less time to get the order into the market. The information is much
more tightly controlled, with Chinese walls in place to stop the cash traders
seeing what the algorithms are trading. All of this ensures that a share
buy-back can be carried out with greater anonymity.
The flip-side is that it could be easier for investors to stealthily build up
shares in a company without alerting the market. Luckily, there are rules
governing just how much stock can be bought before an interest in a company has
to be declared.
Hearing Casper talking enthusiastically about his plans to beat the foreign
exchange markets, it’s easy to believe that the quants with their computers are
taking over the financial markets, pushing out such human concerns as a
company’s underlying profitability.
But Casper cautions against FDs taking a knee-jerk Luddite reaction to
algorithms. “It helps you to carry out your share buy-backs more efficiently and
provides markets with liquidity,” he says. The computers are your friend, not
your enemy, he says.
Over the past three decades, the financial markets have evolved rapidly, making
it hard work for a finance director to keep track of the changes.
All aspects of financial markets have been touched, including the fund
Scroll back a couple of decades and there was pretty much only one way to
manage a share portfolio. A fund manager would buy a share when they thought the
market had underestimated the long-term financial health of a company and, as a
result, its share price was likely to rise.
Deciding whether to buy or sell a share based on an analysis of the underly
ing company is sometimes referred to as discretionary money management – it is
the fund manager who decides when to increase or sell a position.
While there is still a significant proportion of funds which are managed on
this basis, there has been an explosion in the use of quantitative finance over
the past five years.
At is simplest, quantitative finance is a series of trading strategies. Over
time, this has evolved into a different way of managing money. Rather than
making a decision to buy or sell a stock based on the underlying performance of
the company, quants base their investment decision on a statistical analysis of
As the study of statistics has evolved and computational power has allowed
more data to be crunched ever more quickly, quant strategies have rapidly
progressed. They are no longer simply tracker funds that mimic the performance
of an index, but are now complex scientific algorithms to produce market beating
At your discretion
Sophisticated technology means that these complex algorithms decide the buy and
sell orders. They have removed the fund manager’s discretion.
The ability of quant funds to make such high returns over a long time means
there are plenty of investors who now believe it’s better to use statistical
methods rather than fundamental analysis as the way to generate the best returns
from the market.
The lesson for FDs is that they should try to keep pace with the rapidly
evolving technology of high finance to ensure that all shareholders – both quant
and discretionary – have access to the best possible information about the
For more on the predictive power of computers, click