In Norse mythology, Ragnarök refers to future events that ought to predict the death of some major gods. The term was referred to mean the ‘Twilight of the Gods.’ Ardent readers of investment literature would have heard of names such as Peter Lynch, Ray Dalio, Bill Ackman and other fund managers that achieved stellar results. However, can the fund manager, or more generally, the human investor become obsolete in the future? Can we replace the human decision-making processes with mechanical, mathematical based investments? After all, the economy as a field of study, it is full of perfect formulas that suggest how precisely rational we should be when making a decision.
This article will discuss an episode of the possible investment profession’s Ragnarök: are HFT algorithms likely to replace human judgment in making capital allocation decisions? Whether we like it or not, technology is shaping the way we do business and one day it might change the fabric of the global capital markets. However, we will not look to answer the question with an absolute ‘Affirmative’ or ‘Negative’ as this is impossible. Nor will we look to argue for one side or the other.
Instead, we will aim to provide a balanced discussion based on recent research gravitating around the issues and uses of HFT with the hope that this will help the reader in forming an opinion as to whether human reasoning will ever be replaced by the AI brain.
Understanding HFT – The ‘Everlasting’ Definitions
As the World Federation of Exchanges suggests in their HFT guide, there is no singular definition for High Frequency Trading. However, by following the common characteristics of these trading strategies, we can see a pattern emerging: the majority of them are automated trading algorithms that employ variations of market making and statistical data analysis.
The Bank of England, in their February 2015 Working Paper No. 523 – Interactions among high frequency traders, paint HFT as:
“automated computer trades’ interacting at ‘lightning-fast speed with electronic trading platforms’ that has become an ‘important feature of many modern financial markets.”
This gives HFTs a massive speed advantage over human traders because, obviously, computers are much faster at receiving, processing and acting on new information.
The reason as to why it is hard to precisely define HFT, is that it is rooted in the use of these algorithms scaled over different asset classes, pricing models, and market conditions. Nevertheless, automated trading has become an important part of financial markets and represents a good proportion of what is known as the ‘electronic markets.’
Risks and Rewards – The Balance of HFT
The HFT interactions are one of the most important features that can and do impact the stability and functionality of financial markets. There appears to be a high correlation between HFT firms, i.e. a massive sell or massive buy of an asset as a result of a piece of information that triggers that transaction, has been, rightfully so, a big issue associated with algorithmic trading in general (Mary J. White in Enhancing our equity market structure, speech in 2014 accessible here).
For example, the aforementioned Working Paper from the Bank of England focuses on analysing the correlation of HFT algorithms within the UK equity markets. The Paper looks to established whether such correlations (which are synchronised movements) are a danger to market stability. The UK’s central bank determined that the HFT activity is not only correlated across the stocks but also within the stocks (i.e. buying and selling shares from the same company with an algorithm). This means that HFT firms are very aggressive with their buy – sell strategies that can indeed damage market stability (the Flash Crash of 2010 is just one example but in the US).
Another important consideration is the fact that HFT algorithms use market data (especially past market data) to run, suggest and complete trades. Therefore, almost by default, the HFT strategies are limited to this data and those that make predictions or react to new information based on previous market patterns can repeat all the mistakes made by traders in the past.
What about the benefits of using HFT? As the paper was written by Jonathan Brogaard, Terrence Hendershott and Ryan Riordan, High Frequency Trading and Price Discovery, informed high frequency traders play a beneficial role in price efficiency by ‘trading in the opposite direction to transitory pricing errors and in the same direction as future efficient price moves.’ Moreover, they play a beneficial role in creating market liquidity at a reduced transaction cost when compared to traditional market makers (how much liquidity is needed, remains a question deemed for other articles).
Another benefit of algorithmic trading comes from the abovementioned risk – strategy correlation: the improvement of price efficiency as a result of HFT strategies’ correlation in the context of convergence trades, as Kondor P. showed in 2009 in the article published in the Journal of Finance, Risk in Dynamic Arbitrage: Price Effects of Convergence Trading. The difference lies in the triggers of information fed to the algorithm to react and how to do so. Additionally, the regulatory pressures on HFT companies are not as stringent as the ones on banks, brokers, and other market actors – at least not for the moment.
Stock picking – The Human Element
So, will the ‘gods of investment’ be replaced by machines? Whilst we have looked at some of the risks and benefits associated with HFT; we now look at how humans make decisions when investing and we shall see the two processes are almost entirely alien from one other.
First of all, it is important to distinguish that HFT is not used to pick good quality companies. HFT is heavily information-based trading which means that these algorithms react in milliseconds to new information that enters the market (a high volume of trades of low prices are executed this way). Whereas stock picking, or the art of investing, involves a balancing exercise performed by the human brain in assessing quantitative data (numbers, financial ratios, graphs, projections etc.) and qualitative information (the management’s ability, the ‘believe in the system’ and everything that has to do with ‘intuition’).
Secondly, actively managed funds, despite their historical market underperformance (as suggested by legendary investor John C. Bogle in The Clash of the Cultures: Investment vs. Speculation) play a major role in the education of future investors. It is crucial to understand that the markets do not function based on beautifully designed formulas and that the prices of the assets bought and sold almost never reflect their true value (i.e. almost never incorporate correctly the information available because the information is available). Human investors’ experiences and decisions serve as an important basis for understanding how we make decisions when allocating capital, something that no machine or algorithmic trading strategy can teach us. In fact, we are building these HFT strategies based on our understanding of what the markets are and what they are used for.
Finally, we shouldn’t see HFT and stock picking as contradictory approaches to allocating capital but simply as different tools for doing so. Our world, regarding financial engineering but obviously not limited to it, has progressed tremendously: future crashes, economic downturns, bull and bear markets and rogue trades will take place, and we cannot entirely prevent them. However, as we develop new ways of viewing and engaging with the financial markets, most of them based on digital and IT infrastructures, we ought to be concerned with questions such as: Do our investments and trades create value or do they simply create money? Do our capital allocation decisions and practices endanger the financial markets at large or do they help to facilitate their function? These are some of the ideas that we should meditate upon regardless of whether we are in the HFT business or the stock picking camp.