High-frequency trading explained: why has it decreased?

‘People no longer are responsible for what happens in the market, because computers make all the decisions,’ – Michael Lewis, author of Flash Boys.

More than half of all equities traded in the US is done not by humans but by super computers capable of placing millions of orders each day and gaining advantage through moving milliseconds before the competition. This high-frequency trading has seen market makers and the largest players use algorithms and data to make money from placing vast amounts of orders to earn wafer thin margins.

But these margins have become even slimmer and the opportunity has dwindled: revenue last year was around 86% lower than it was when high-frequency trading was at its peak less than a decade ago. With pressure continuing to grow consolidation has started to take hold of the sector as high-frequency traders look to fend off tougher conditions.

We have a look at what high-frequency trading is and why it has declined.

What is high-frequency trading?

It comes to down harnessing the power of technology to gain advantages whilst trading. High-frequency trading sees large organisations such as investment banks and hedge funds use automated trading platforms that, using algorithms, are able to track numerous financial markets and execute vast amounts of orders.

Learn more about algorithmic trading and how it works

If programmed correctly, high-frequency trading offers an obvious advantage to those institutions that have access. The highly powerful computers can spot new trends across global financial markets and act automatically before the rest of the market has had a chance to even identify the trend, let alone trade it.

Read more about automated trading and how it works

The generally accepted characteristics of high-frequency trading and the firms that do it are:

1. Deals in extremely high number of deals
2. Orders are rapidly cancelled
3. Holds positions for very short periods of time
4. Holds no positions at the end of each trading day
5. Earns wafer thin margins per trade
6. Use of data feeds and proximity services
7. Proprietary trading (when a bank etc. invests for its own gain rather than on behalf of clients)

Why does high-frequency trading exist?

High-frequency trading allows large institutions to gain a small but notable advantage in return for providing vast amounts of liquidity into markets. The millions of orders that can be placed by high-frequency trading systems means those using them are lubricating the market and, in return, they are able to increase profits on their advantageous trades and obtain more favourable spreads.

The nature of high-frequency trading satisfies both sides. Institutions can gain an advantage but one that is based on volume. The returns per transaction are tiny and therefore a huge number of trades must be completed to truly benefit which, in turn, ensures enough liquidity is being pumped into the markets. Firms have two direct income streams: one from earning the spread for supplying liquidity and another through the discounted transaction fees that trading venues provide to make their markets more attractive to high-frequency traders.

High-frequency trading: spreads and liquidity

Spreads and liquidity go hand-in-hand. Markets with high activity levels offer smaller spreads while those with lower trading volumes tend to offer larger spreads: with the spread being the difference in price between the buy (offer) and sell (bid) prices quoted for an asset.

The foreign exchange market, for example, sees over $5 trillion of currency traded each and every day to hold the title as the most liquid market in the world and it is this staggering volume that creates small spreads that only offer material profit opportunities if they are traded in large volumes. It is this reason why many choose to use leverage in markets with high liquidity such as forex, so volumes are maximised in order to take more substantial positions that otherwise might not be worthwhile. For less liquid markets such as small-cap stocks the spreads on offer are typically much larger.

What is leverage?

Is high-frequency trading ethical?

Giving large institutions an advantage over smaller organisations and retail investors raises obvious questions about the ethics and fairness of high-frequency trading, and rightly raises the argument that it doesn’t help promote a level playing field. As well as competing with one another retail investors have to compete with an algorithm that is far superior than human trading.

Some also argue that the liquidity provided by high-frequency traders – the service they provide in return for gaining the advantages – is not as effective as its supposed to be because the speed at which high-frequency trading operates can see money flow in and out of a market within a blink of an eye, preventing other investors from benefiting from it.

Benefits and drawbacks of high-frequency trading

How significant are high-frequency trading volumes?

The US has always been the main hub for high-frequency trading, which has accounted for at least half of all the volume within the US equity market every single year since 2008. Volumes peaked at 60% in 2009 but, as the financial crisis took its toll, the share of high-frequency trading began to decline before stalling at 50% for three consecutive years until 2016 when its share began to climb again.

_{Source: TABB Group, Deutsche Bank, ResearchGate}

High-frequency trading has not been as dominant in Europe – although still very significant – and the US was much quicker to adopt it. High-frequency trading only properly started to emerge in Europe in 2006 when the method already accounted for around 25% of US equity volumes. There has been a strong correlation between high-frequency trading volumes on both side of the Atlantic: European volumes peaked a year after the US in 2010 and has since followed the same pattern.

Why has high-frequency trading revenue collapsed?

The revenue generated through high-frequency trading peaked in the same year as volumes but the decline post-2009 was far more aggressive. In eight years high-frequency traders in the US have seen revenue from the equity markets collapse from a peak of $7.2 billion to below $1 billion in 2017 for the first time since the financial crash, according to estimates from consultancy firm TABB Group.

_{Source: TABB Group, Deutsche Bank}

There are numerous reasons why the rewards of this practice have dwindled over the last decade or so. In a nutshell: increased competition, higher costs and low volatility have all played their part. High-frequency trading firms have been squeezed from both a revenue and a cost perspective, and the effect of that is exacerbated considering millions of trades need to be completed daily for the practice to be worthwhile. Vikas Shah, an investment banker at Rosenblatt Securities, told the Financial Times earlier this year that high-frequency traders have two raw materials they need to effectively operate: volumes and volatility.

The speed at which high-frequency trading operates means every nanosecond counts. But algorithm trading comes down to a zero-sum game based on how fast current technology can go. Once everyone is at the same speed the advantages high-frequency trading offers disappears. Not only does this mean every slight update is essential for the already highly expensive equipment but that the value of the vital data needed for its algorithms to work has risen exponentially. This data has to be acquired from the trading venues which, noticing the sector’s surging income, they used to their advantage.

'Co-location services', as they are known, allows a company to rent space in the trading venue’s data centre or server to secure a direct link to the swathe of price movements and other data as it emerges. According to Deutsche Bank, the co-location fees charged by major exchanges 'doubled or tripled' between 2010 and 2015. Ironically, when volumes fall exchanges lean on other sources of revenue such as selling data, but the higher cost of data has been one of the reasons why high-frequency trading volumes have dropped.

The relentless appetite to gain even the slightest edge over the competition has even pushed companies to move their physical location closer to the data servers because, apparently, the fraction of time gained by not having to send the information as far through the Internet is that valuable. This has spurred on a new breed of infrastructure provider aiming to connect trading venues and high-frequency traders with ever-faster cabling. Regardless of what tact they are using, the cost of high-frequency trading has undoubtedly risen and made it a less attractive option.

High-frequency trading and dark pools

Considering the importance of data for high-frequency trading and the fact the cost of such data is rising the role of dark pools is significant. These are private exchanges where institutional investors trade large volumes with one another without having to disclose the details of the deal to the wider market. This also means the transactions conducted in dark pools bypasses the servers feeding the data used by the algorithms established by high-frequency traders.

Dark pools are controversial. On one hand there is an argument in favour for them as the biggest players can trade large volumes without upsetting or disturbing the wider financial markets. On the other is the argument that they provide a way for corporate giants to deal amongst themselves while leaving everyone else in the dark.

These private exchanges are nothing new. Dark pools have been around since the 1960s and although data from these exchanges is slim it is thought the volume being traded has grown while the level of high-frequency trading on public markets has fallen. This is because the ability to trade large volumes on dark pools without causing severe price movements in the market means high-frequency traders have less opportunity to conduct larger trades on public markets, which in turn has put more attention on lower-volume deals which high-frequency trading is not designed for. Previous ‘flash crashes’ or sharp price movements caused by high-frequency trading has only glistened the appeal of dark pools.

Consolidation of high-frequency traders

The growing pressure on high-frequency trading has led to consolidation within the sector as companies combine to fend off higher costs and tougher market conditions. While the majority of high-frequency traders are private there are some publicly-listed companies involved in the sector such as Citadel Group, Flow Traders and Virtu Financial.

Virtu listed in 2015 and last year bought peer KCG Holdings in a $1.4 billion deal that saw it account for about one in every five trades conducted on the US equity market, and has recently been reported to be eyeing Investment Technology Group.

Other recent deals in this space saw one of the largest high-frequency traders, DRW Holdings, buy RGM Advisers last year and two further rivals merged after Hudson River Trading acquired Sun Trading after the latter was put up for sale in 2017 as margins came under pressure and its competitive edge in terms of speed was lost.

High-frequency trading regulation to continue evolving

‘Reg NMS was intended to create equality of opportunity in the US stock market. Instead it institutionalised a more pernicious inequality. A small class of insiders with the resources to create speed were now allowed to preview the market and trade on what they had seen,’ – Lewis.

The most substantial piece of regulation considered to have spurred on high-frequency trading from 2005 onwards was the introduction of the Regulation National Market System (Reg NMS) in the US. This regulation is what gave traders the insight into the strategies of other investors in the hope that, in times of crisis or during downturns, trading would continue rather than result in non-communicative brokers avoiding taking sell orders as they had done in the 1987 crash.

The International Financial Law Review highlights one rather notable aspect of Reg NMS that meant all orders that were placed had to be executed at the best price regardless of what exchange it is on, thus allowing high-frequency traders to spot trends in one exchange before rushing to capitalise by placing orders on another exchange before the effect has had a chance to ripple through. While it was meant to provide a more transparent and level playing field between the largest players in the financial market, everyone else was put at a disadvantage.

Things have been tightened since, with MIFID II in Europe and FINRA in the US both including rules on algorithm trading. The London School of Economics and Political Science states a major problem with regulating high-frequency trading is defining exactly what it is. While there are generally accepted characteristics there is no universally accepted definition.

Still, MIFID II implemented new rules requiring high-frequency traders to gain authorisation from market authorities and required better record-keeping as part of wider attempts to stamp out any abuse. However, there is widespread acceptance that there is much further to go in regulating the sector.