As every year ends there is usually an assessment of who were the winners and losers in the world of global financial trading and investments. We want to know who made the best bets and invested in the right companies. And we want to know who had that particular insight which enabled them to realise that, although everyone was optimistic about a particular stock, the fundamentals did not add up, and as a result decided to short the stock and invest in another company everyone had underrated.
But this insight seems to be moving away from human beings to computerised models. So is the role of discretionary investment waning at a time when we have powerful computers and software, with an increasing ability to decipher meaning from vast amounts of changing data and come to appropriate decisions?
We are in an age where there is a vast amount of real time information from various sources in several parts of the world. This information has an effect on the value of stocks, bonds and other investment instruments. The speed and volume of information leads one to consider whether a machine using algorithms is much better at investment analysis and coming to the best decision than smart traders with several years of experience. And at a time when automated systems are able to process so much data and give us accurate answers, is there a lot of value in common sense and fundamental economic analysis carried out by an individual?
When making investment decisions, computers are able to track complicated market patterns and observe and analyse data from digital sources like news stories and social media. They carry out these tasks without most human weaknesses, for example, being depressed, tired or angry. They also do it without subconscious factors that affect judgment such as emotional connection, historical precedent or sticking to a view it that seems correct because it is theirs. The most important of these factors is cognitive and confirmation bias. When making investment decisions our judgement is affected by factors based on our experience. For example if a person invested in emerging markets and lost money, or you invested in the technology sector just before the dotcom crash. This affects how a person will assess an investment opportunity based on their experience; sometimes they even have flashbacks which will affect their objective assessment of the investment opportunity. With algorithmic trading all of this is absent and a clear unbiased view of the stock or bond can then be carried out giving an objective view.
Another important factor is confirmation bias, as here the person already has a view and will search for factors that support that view. Their perception and judgement is skewed because they are already have a view, whereby they carry out research but only place emphasis on the facts that confirm what they already believe.
This allows computers and other automated systems to interpret the market with more objective and precise results. These automated systems also have the capacity to predict how stocks will perform in future, new fashion trends and election results. Some say computers and their algorithms will also put the specialist advisers in these sectors out of work.
But others do not believe that automated technological systems can ever predict the future, no matter how much data they gather and analyse accurately. A human carrying out discretionary analysis can assess obscure values and factors that cannot be quantified accurately or numerically. Most importantly, humans can use their intuitive sense to assess a company’s management, its leaders’ psychological behaviour, team spirit and work ethic.
Human beings can predict the irrational behaviour of customers in trends and news, all based on intuition or gut feeling. Be that as it may, in global financial markets, analysing securities will never be the same because the amount of data available has increased dramatically, and analysts using automated systems have an advantage.
Their advantage comes from algorithms. Algorithms are rules and instructions used to inform computers on how to find and interpret data, and have become an indelible part of our lives. For example, I am using one to spell check as I write this article, and I will use one - my GPS - when I drive to work tomorrow morning. We are not far from a time when algorithms will drive us to our destination without us even touching the steering wheel.
In the digital global market algorithms allow us to assess and process the vast megabytes of data emanating from different sources. This is used to carry out economic forecasts and trades. Most impressive is the ability to analyse news in real time: here algorithmic programmes observe customers, companies and investors while they are conducting their activities and compare them with their competitors. These computerised systems highlight trends that could have consequences in the market.
Some say that this will bring about the demise of human analysts, as the computerised algorithm can discover patterns of behaviour and predict what will happen in real time. It carries this out much faster and more accurately than any human judgment or observation.
But those that hold a contrary view say that human intelligence is still in control because algorithms, like economic models, are put together by human beings with a particular perspective of the world and a prescribed objective. Human beings are able to be more flexible, and can consciously or subconsciously factor in other human emotions and irrational sentiments before making decisions. A human research analyst is able to spot the changeable factors that computerised systems need a while to carry out. However, they cannot synthesise data as systematically as computers.
Computers using algorithms will always carry out the same task much more successfully and predict returns or results in a more precise way than any human. However, when an algorithmic model has been built based on certain premises we need human beings to intervene when unplanned or unprecedented events occur out of the blue. This includes events such as the attack on the World Trade Center in New York on 11 September 2001 or an earthquake.
In financial market trading, this has become a very important distinction in that computerised systems can be agnostic of all asset classes and time horizons, as well as being very diversified. They can identify themes using long-term macro variables, which adjust to the ups and downs in financial markets opportunistically. The algorithms deal with periods where there is transition, an event risk like the departure of a company’s chief executive officer, changes in fiscal and monetary policy, or an event which triggers a rise in market prices. In such circumstances the algorithms can convert most of their allocations to cash for safety. The computerised trading model avoids emotional bias and focuses on the facts while assessing data and making allocations.
These computerised models are able to identify themes and adjust accordingly without much reliance on market timing, as this normally requires several technical indicators. Such indicators are very useful while spotting trends but have to allow a wide margin for error and adjust accordingly as things change in financial markets. Simply spotting trends is not enough: the model needs to be robust enough to capture the risk premium in several market conditions over the short and long-term. These computerised trading models use indicators to:
- ascertain the best time to buy
- assess how long to hold
- determine when to sell.
Fund managers charge fees on the basis that they are able to generate better returns by carrying out superior research, reduce risk and generate higher returns than most indices. Those that aim to follow the index are described as passive investors. Fund managers invest actively by carefully selecting portfolios that reflect their investment thesis and approach to reducing risk. Some now also use algorithms to search for alpha (above index returns) by trading in several markets with a trend-based strategy. This allows them to go short, which means that the stock is overvalued and will reduce in price; or long, when the market is undervalued and will increase in price.
Most fund managers aim to maximise returns and preserve capital, thus facing the dilemma of risk versus returns, although risk does not have a specific definition.
For most investors, when considering risk, one of the first things that springs to mind is volatility, so they consider how the level of contingency and variability will affect the expected return on an asset.
However, there are other types of risk, such as credit, counter party, liquidity, event and market risk. This is why some algorithmic systems have risk budgets which allow for measurement of dollar at risk and ascertain how much leverage should be applied. These models adjust for market microstructure risk by isolating and identifying the risk premium in asset classes more accurately. The amount of risk is capped to account for limited upsides in certain conditions.
These algorithmic models should not be overly reliant on prediction. They should be able to operate under most market conditions, follow trends when necessary and go against these trends when the situation arises. Most importantly, they should be able to bounce back strongly in difficult market conditions.
A human discretionary investment manager has a much better ability to bounce back. However, many computerised models carry out reasonable assessment called back testing before trades are executed, based on research and a simulated market environment. These checks serve as insurance against unforeseen events and system errors.
Quantitative investing is the scientific approach used to rationalise the activity of investment management after constructing a view of how the market functions, taking monetary and fiscal conditions into consideration. Fund managers empirically test their investment strategies, just like process engineering. Examples include Harry Markowitz, who developed a model that allowed diversification by attempting to find out the correlation between various asset classes, and Bill Sharpe, who developed an approach to evaluating the forward looking trade-off between risk and expected return, which he called the Capital Market Line Model.
The most important of these is the efficiency market hypothesis. This attempts to explain the idea that it is impossible to beat the market because the price of stocks will return to their fair value in the long run. Proponents of this idea argue that attempting to predict trends using technical or fundamental analysis is not very useful in the long-term.
Fund managers also aim to provide portable alpha for their clients. Alpha is the return higher than the market average or benchmark a manager generates via his or her skill in making the appropriate allocations. Another factor is market risk, and the correlation the manager has to market risk, which is known as beta. If this is low we give the fund manager higher marks because his or her risk is lower than that of the market they are trading in. Portable alpha is when the fund manager invests in a different market from the one he or she is generating beta from. Many investors yearn for this portable beta.
To allay the public’s fears about programme trading, the Brady Commission in the US brought about a system called circuit breakers. Circuit breakers bring a halt to trading if indices like the Dow Jones fall below a certain level, say 20%. This provides some safety by preventing computerised systems from driving equities too low too fast, and removing short-term market volatility. So in this case, human intervention plays a crucial part.
Applying a quantitative computerised approach to investing allows us have a much clearer understanding of the process that occurs during investment, as it eliminates cognitive, emotional and confirmation bias. This allows us as investors to attain a better perception and a more precise result. This, however, does not replace fundamental analysis, economic assessment, common sense and, most importantly, a human’s intuitive understanding of other human beings in the market.
Francis is a founding principal of Majlis Partners and managing director of the London office. He is responsible for Business development, strategic initiatives, deal sourcing, manager sourcing and various marketing and research functions across the firm. Francis Akpata has more than 15 years of experience in international marketing and business development Mr. Akpata has systematically built-up a network of clients/investors in the GCC countries (Bahrain, Kuwait, Oman, Qatar, Saudi Arabia and United Arab Emirates) the U.S. and Europe (Italy, Monaco, UK, Switzerland and Scandinavia). In these markets, he provides tailored but workable solutions for the distribution/marketing of products, services, public relations, and investor relations. Since 2008, he has worked specifically to provide solutions that reduce the risk for Investors and assist companies with developing their business in the hedge funds and private equity sector. Francis was head of investor relations at Acropolis Capital that is a family office that has been investing in hedge funds and private equity for over ten years. At Acropolis Mr. Akpata’s role was to distribute their in-house products to institutional and private clients (Family offices and HNWI) in Europe and the Middle-East. Prior to that Francis was at Mellon Global Investments (Now Bank of New York Mellon) in London. Here he developed marketing strategies for the sales teams and facilitated the development alternative products.
Majlis Partners is a merchant Advisory firm that helps firms and institutions develop partnerships and alliances in the GCC and Europe and other parts of the world. Majlis has an open architecture platform focused on show casing best-of-breed companies and institutions targeting sovereigns, family offices, and institutions in the GCC, Europe and other parts of the world. For more information, please visit www.majlispartners.co.uk