High-Frequency Trading (HFT)

What is high-frequency trading (HFT)?

High-frequency trading (HFT) is a specialized form of algorithmic trading that involves executing a large number of trades at extremely high speeds. It uses powerful computers, ultra-fast data feeds, and complex algorithms to scan markets, identify tiny price discrepancies, and automatically execute trades when predefined conditions are met.

HFT trades are placed and cancelled in fractions of a second. The goal is not to capture large price moves, but to generate small profits – often just fractions of a cent per trade – across thousands or millions of transactions. If an algorithm detects a brief pricing inefficiency or arbitrage opportunity, it can trigger hundreds of buy or sell orders in an instant.

Because of the speed, complexity, and infrastructure required, HFT is mostly used by large institutional investors like investment banks and hedge funds.

History & development of high-frequency trading

HFT has been in slow development since the early 1980s, when electronic trading first started replacing traditional floor-based trading. In the early 2000s, trades could be executed in several seconds, but HFT was still relatively unknown and accounted for less than 10% of equity orders. 

Between 2005 and 2009, HFT trading volume grew by about 164%, as execution speeds dropped to milliseconds and even microseconds. It became even more popular after the 2008 financial crisis, when exchanges offered companies incentives for adding market liquidity. For example, the New York Stock Exchange (NYSE) offered supplemental liquidity providers (SLPs) a fee or rebate in exchange for adding competition and liquidity.

Since then, HFT trading has become a mainstay of financial markets, accounting for about 50% of US equity trading volume.  

How HFT works

HFT involves using automated systems to continuously monitor market movements, price changes, trading volume, and order flow across multiple instruments and venues. When a predefined condition is met (i.e. a brief pricing imbalance or arbitrage opportunity), the algorithm automatically executes a trade without any manual intervention. This allows traders to respond to short-lived market opportunities that may not have been spotted otherwise.

Here’s a look at the high-frequency trading process, step-by-step:

1. Market data ingestion

HFT systems receive a constant stream of real-time market data directly from exchanges. This data includes prices, trade volumes, and order book information at different levels (Level 1 and Level 2 data). This provides a fast-paced, up-to-date view of current supply and demand, down to the millisecond.

2. Microsecond-level analysis

Once the data is received, the algorithms begin analyzing it to find very small inefficiencies. This could be pricing discrepancies between markets, order imbalances, or temporary mispricings. It takes just milliseconds or microseconds to complete the analysis.

3. Automated decision-making

The system then uses predefined rules and statistical models to decide whether to place a trade. These decisions are fully automated and follow strict logic, so there’s no emotional decision-making or human involvement.

4. Trade execution

If the conditions are met, orders are sent to the market almost instantly. HFT firms use ultra-low-latency infrastructure to execute trades extremely quickly. Some even place servers close to exchange data centers to benefit from proximity (called co-location).

5. Order modification or cancellation

In practice, many HFT orders are adjusted or cancelled before they’re filled. Markets move quickly, and if conditions change even just a little bit, the algorithm might withdraw or update an order immediately.

Low-latency infrastructure in HFT

Speed is everything in high-frequency trading. To reduce delays as much as possible, firms invest heavily in infrastructure including:

• Co-location: Placing trading servers physically close to exchange servers can help reduce transmission time
• FPGA & custom hardware: These specialized processors can analyze data faster than standard computers
• High-speed networks: Microwave networks can transmit data with minimal delay, faster than fiber optics
• Smart order routing: Automatically select the best trading venue for execution
• Low-latency APIs: Interact with exchanges at high speeds.

High-frequency trading strategies 

Market making

Market making involves continuously placing buy and sell limit orders in order to earn the bid-ask spread. The algorithm places buy orders slightly below the market price and sell orders slightly above it. When both sides are filled, the trader captures a small profit. The strategy is repeated at high speeds and volumes to generate consistent returns.

Statistical arbitrage

Statistical arbitrage strategies aim to profit from temporary deviations in historically stable relationships between assets. These relationships might exist between stocks, bonds, futures, foreign exchange, and other instruments.

Algorithms monitor these relationships in real time and execute trades when prices diverge from their expected patterns. Positions are closed once prices revert to normal.

Index arbitrage

Index arbitrage focuses on index tracker funds that buy and sell large volumes of securities when their compositions change. Traders use HFT systems to anticipate these moves before the fund does, allowing them to buy up ahead of the trackers and sell them later at a profit.

News-based trading

News-based HFT strategies analyze real-time news feeds, economic releases, social media, and corporate announcements to identify trading opportunities and execute trades before human traders have time to react. 

Low-latency arbitrage strategies

Low-latency arbitrage strategies aim to exploit tiny price differences in the same asset trading on different markets simultaneously. The idea is to buy the security at a lower price in one market and sell at a slightly higher price in another, often within milliseconds.

Advantages of HFT

The advantages of high-frequency trading include faster trade execution, improved market liquidity, and tighter bid-ask spreads.

Faster and more efficient trade execution

HFT systems can analyze market data and execute trades in milliseconds or microseconds, allowing institutions to place and manage a large number of trades in an extremely short period of time. This level of efficiency is difficult – or even impossible – to achieve when managing significant order flow through manual or slower algorithmic trading.

Improved market liquidity

High-frequency trading firms often act as market makers by continuously placing buy and sell orders. This increases the availability of counterparties in the market and makes it easier for other participants to enter and exit positions.

Narrower bid-ask spreads

HFT also helps tighten bid-ask spreads, making the difference between buying and selling prices smaller. This reduces trading costs for all market participants, including retail traders.

Enhanced price discovery

HFT trading systems react to new information almost instantly. When there are temporary price misalignments across markets or venues, HFT algorithms exploit these discrepancies and help prices adjust to a more fair value. This improves price discovery and overall market efficiency. 

Ability to profit from small price movements

HFT strategies are designed to capture very small price changes repeatedly. Even a tiny profit per trade can generate meaningful returns when executed at high volumes.

HFT risks and controversy

Despite its advantages, HFT has been subject to criticism when it comes to:

Market volatility

One of the main criticisms of HFT is that it can contribute to market volatility. Because trading decisions are made in milliseconds, algorithms can react simultaneously to the same signals, triggering sudden and sharp price movements without any fundamental change in market conditions.

This risk presented itself in 2010, during what’s known as the Flash Crash. On May 6, the Dow Jones Industrial Average (DJIA) experienced its largest intraday drop, falling 1,000 points and 10% within 20 minutes before quickly recovering. Investigations later blamed the crash on a massive order that triggered a sell-off.

Concerns about ‘ghost liquidity’

Another criticism of HFT is that it creates 'ghost liquidity'. Even though high-frequency traders post large numbers of buy and sell orders, the liquidity is not considered 'real' because the securities are only held for a few seconds. By the time a regular investor places an order, the apparent liquidity created by HFT has likely already disappeared.

So although HFT gives the impression of deep liquidity, this liquidity often disappears before other traders even have a chance to trade it.

Unequal market access

HFT requires expensive infrastructure and specialized hardware to reduce latency, raising concerns that it gives large institutions an unfair advantage over retail traders and smaller firms that can’t compete on speed. This means HFT firms can receive and act on price updates faster than others, creating an uneven playing field in the market.

Algorithmic errors

High-frequency trading depends on using software, leaving it vulnerable to coding errors or faulty logic. In 2012, for example, Knight Capital Group suffered a $440 million loss in just 45 minutes due to a malfunctioning algorithm triggered by a failed software deployment.

Market manipulation practices

Most high-frequency trading strategies are legal, but there have been situations where HFT firms engaged in illegal practices like spoofing, layering, and quote stuffing. These strategies involve placing large numbers of orders with no intention of execution, just to mislead other market participants about supply and demand. HFT firms who engage in these strategies can be subject to fines and enforcement actions. 

How is high-frequency trading regulated?

In the U.S., regulatory bodies like the Securities and Exchange Commission (SEC) and Commodity Futures Trading Commission (CFTC) have introduced rules designed to reduce systemic risk and prevent abusive trading practices amongst HFT firms. This includes issuing significant fines for inadequate risk management, market manipulation, and failing to disclose order types properly.

For example, Knight Capital was fined $12 million in October 2013 following its trading malfunction. Regulators believe the firm had violated the SEC’s market access rule and ignored dozens of error messages before the system eventually sent millions of erroneous orders to the market. In 2014, another HFT firm, Latour Trading LLC pais a $16 million penalty after the SEC found it used faulty calculations to buy and sell stocks without holding enough capital. 

In Europe, MiFID II introduced stricter regulatory oversight to HFT and algorithmic trading by requiring authorization, detailed record-keeping, strong internal controls, and technical standards to prevent market abuse.

HFT vs. algorithmic trading

High-frequency trading is a subset of algorithmic trading. Here’s a look at the main differences between the two:

HFT in cryptocurrency markets

High-frequency trading also exists in cryptocurrency markets. The same mechanics are used as with other markets, like equities. Cryptocurrency HFT systems use algorithms to analyze real-time crypto data and execute a large volume of trades within seconds.