Experiments on the effects of trading algorithms on financial markets

Algorithmic trading, where computers rather than humans execute trades within milliseconds of receiving information, has been on the rise for years. Some authors estimate that 70% of trades in the US are executed by algorithms (Swinburne 2010), with others going up to 85% (Glantz & Kissel 2013). Algorithmic trading has thus had a significant impact on all developed financial markets, but its effects are not yet well recognised; an improved understanding of this relatively new and shifting phenomenon is essential to inform policy debates, such as whether algorithmic trading needs to comply with stricter rules to prevent excess volatility or the formation of asset bubbles.

The aim of this project is to answer fundamental questions on the effects of algorithmic trading on financial markets; specifically, whether algorithms trained by state-of-the-art machine learning techniques increase or decrease the likelihood of asset bubbles, whether they increase or decrease volatility and risk in markets, and indeed how such algorithms look like in the first place. That latter question is simple yet not easy to answer, because successful trading algorithms are valuable and therefore kept confidential for fear of them being copied, exploited, or regulated. Another difficulty is that causal effects are notoriously difficult to establish with field data from financial markets, in part because it is typically not even determinable which trades are executed by algorithms in anonymous markets.

For these reasons, I will answer these questions with laboratory experiments, where I let human traders and algorithmic traders interact in small lab markets. The advantage of the method is that causal effects can be convincingly established: A control group with only human traders can be compared to a treatment group with algorithmic traders along various outcome measures such as bubble frequency, volatility/risk, or liquidity/trading speed, all else equal. Moreover, every action is observable in the lab, so specific trades can be classified as being initiated by algorithms or humans. It also allows me to investigate which kinds of algorithms evolve when learning against human traders, or against other algorithms in the market.

One class of market used in the lab will let subjects/algorithms trade multiple units of a finanical asset. This asset pays out a dividend to the holder at the end of each of 15 rounds. Because there are fewer dividend payoffs remaining in later rounds, the fundamental value of the asset is decreasing over time. Yet in previous experiments with human traders, financial bubbles frequently arose where prices increased dramatically despite the decreasing asset value, which is the bubble pattern also observed in stock or housing markets. This experimental market setting is therefore useful to study bubble formation and causes of bubbles, but so far there is no research on the effect of algorithmic trading in this context.

In another class of experimental market, I will introduce news about the value of assets during financial market trading, and observe how traders and algorithms respond to such news shocks. In particular, the main question is whether a sufficient number of trend-following algorithms can lead to drastic drops or increases in prices, thus increasing volatility and risk in markets, or whether algorithms may actually stabilise markets.

The research findings will inform policy formulation and have significant impact potential. Determining the role that algorithms play in market events such as financial bubbles and 'flash crashes', where prices drop swiftly without apparent cause or news, only to bounce back shortly afterwards, will assist the finance and regulatory sector in deciding whether invention or regulatory changes are required. The experiments are also a good and inexpensive tool to test the effectiveness of new policy measures or regulations.

The proposed research will primarily benefit financial regulators and policy makers, because the research questions have large policy implications. In the UK, these primary beneficiaries are the Bank of England and the Financial Conduct Authority (see FCA 2016 for some of their work on the issue). Regulators around the world face the same questions regarding algorithmic trading, and policy/regulatory responses depend on the answers. Hence, international regulators, like the European Central Bank, the Federal Reserves in the US as well as the SEC, or the Bundesbank and BaFin in Germany (financial market supervisory authority), to name just a few, stand to benefit from this research.

Problems on financial markets can have severe effects on the real economy, as the financial crisis of 2007/08 has shown. My research aims to address two of the most pressing policy questions related to algorithmic trading. First, does algorithmic trading lead to better market functioning and more liquidity, or to adverse effects like more volatility or "flash crashes", thus destabilising markets? Ultimately, this is an empirical question. Opponents of algorithmic trading argue it can cause large and swift drops in financial prices without apparent reason, as a market populated with many trend following algorithms sets a cascade of selling in motion. Proponents, on the other hand, argue the algorithms can decrease human error, and in the case of high frequency trading, actually add liquidity to the market, thus making it easier for others to find trading partners. Since research based on field data has limitations, the policy debate is in dire need of answers to this question. If it turns out that algorithmic trading - or a critical mass of it - has adverse effects as the opponents claim, then regulatory action may be called for. Such regulatory action might include the regulation of exchanges, to provide more of a level playing field between algorithmic traders and other traders, or the direct regulation of algorithmic trading aimed at preventing market destabilising sell cascades. My approach of using experiments to address this question is also a fruitful tool for research on policy responses: Experiments can be used to answer which rules and regulations would minimise adverse effects without outright prohibitions on algorithmic trading.

Second, do algorithms contribute to or ameliorate financial bubbles? There has been a longstanding debate among central bankers (e.g., see former US chairman Bernanke 2002) whether central banks should intervene in markets to counteract the forming of bubbles. Depending on whether algorithms contribute to the problem or help, new policy approaches could be aimed at algorithmic trading specifically, which was not a viable strategy in 2002 when algorithms were not as prevalent and the technology not as mature.

A second class of non-academic beneficiaries are financial exchanges. While regulators carry the force of law, exchanges like the London Stock Exchange can make their own rules for trading on their own platform. For example, exchanges could choose to execute trading orders not strictly in the order of arrival, but randomise the order within a small time interval, so that the fastest algorithms or traders do not always get served first. The FT (2012) makes a similar point by suggesting the delay of every order by a randomly determined small time interval. Some exchanges in the US have taken such steps (FT 2019). These examples show that there is some pressure on exchanges to react, and my research can help to inform or justify such trading rule changes which will affect most modern financial markets, and by extension their real economies. Moreover, exchanges are interested in maximising trade, so they are interested in the question whether algorithmic trading crowds out human trading, and in possible measures to prevent crowding out. My experiments directly address this question as well.