Networked markets: the evolution of high-frequency trading (HFT)
By Lucy Thompson, on 9 November 2022
Financial markets have undergone a deep reorganisation in the last 20 years. A mixture of technological innovation and regulatory constraints has promoted the diffusion of market fragmentation and high-frequency trading. In this blog, IFT PhD student Zihao Liu reports on the inaugural seminar in IFT’s Agora Seminar Series – “High frequency trading and networked markets” with Professor Rosario Nunzio Mantegna.
Due to the high-speed development of FinTech and technical innovations in recent years, the operation of the global financial market has changed beyond recognition. Ever more market participants are beginning to use powerful computer algorithms to execute and complete orders in mere microseconds. The new stock market has changed the traditional ecology of market participants and market professionals.
With the development of strategic trading decisions based on high-frequency trading, the fragmentation of markets has occurred. Contemporary stock markets are now “networked markets” where liquidity provision of market members has statistically detectable preferences or avoidances.
Professor Rosario Nunzio Mantegna’s research investigates this evolution in two highly studied European market venues – London Stock Exchange (LSE) and Nasdaq Stockholm – during three observation periods:
- 2004 to 2006, when high-frequency trading was limited
- 2010 to 2011, as high-frequency trading expanded
- And 2018, as high-frequency trading was further developed
Prof. Mantegna began by identifying high-frequency traders (HFTs). There are two categories of market members (MMs) in the networked market. Based on the definition in this research, HFTs are professional traders able to use high speed in the generation, routing, and execution of orders. Non-HFTs are traditional traders through market makers in institutionalised figures. Today, in most settings, market making is not institutionalised and the amount of transactions performed by HFTs is estimated to be around 50% in most markets.
People have different opinions on these arising HFTs. Some think they improve the market efficiency by decreasing the transaction cost. Others believe their competition will deteriorate liquidity provision. Based on the networked market structure, during market activities, HFTs typically switch between market makers to provide liquidity and back runner to take the liquidity. There are four possible trading pairs among HFTs and non-HFTs, as the counterparty can be either one of them. A random realisation of configuration model is used to show the statistical evidence of the networked structure of transactions occurring between each of these pairs.
(Figures from Rosaria Mantegna’s paper on High Frequency Trading and Networked Markets)
Overexpressed and underexpressed interactions
These are two categories of transactions performed by MMs. Based on the complex network theory, the methodology of statistically validated networks is used in the research to compute the probability of finding a number of transactions larger than or equal to the successful occurred transaction for each MM pair in the market. The probability is approximated by a P-value to test the null hypothesis. Then the number of overexpressed and underexpressed transactions can be found as they interact more or less than the expected probability from the model. Based on the investigation, both these two kinds of transactions are increasing over time. There is also a positive relationship between the trading speed and these detected transactions. It means that more HFTs are joining the market as the market evolves between 2004 and 2018.
After the directed couples of MMs are validated by the network model, the persistence of these transactions is also tested. In the overexpression transactions, there are more persistence in the mixed trading pairs of HFT and Non-HFT. While for underexpression transactions, the persistence exists only in the pairs of HFTs or Non-HFTs. It means that in these two categories of transactions, HFTs may execute different strategic trading decisions to establish preferential or avoided trading relationships with specific MMs.
Finally – is a networked structure the best solution?
HFTs increase trading speed of all categories of investors and provide liquidity in normal market conditions. However, Prof Mantegna’s research concludes that this is in fact suboptimal. The liquidity of HFTs is not provided in a way proportional to the trading interests of investors acting in the market. HFTs are competitive, as they can process the accessible information fast which is not feasible to a large number of market members. But this means there is still a question mark around financial inclusion and fairness, as high-frequency trading may be an unfair advantage for large firms against smaller investors.