High-frequency trading (HFT) is a trading strategy that utilizes automated trading systems to capture and profit from small market fluctuations within extremely short timeframes. For instance, traders can exploit tiny price differentials between the buy and sell prices of a security or seek out small price discrepancies across different exchanges for a particular stock. Due to the high speed involved in these trades, some trading firms even position their server clusters in close proximity to exchange servers to minimize the time it takes for trade orders to travel through the fiber-optic cables.

HFT typically relies on computer algorithms to execute a large volume of rapid securities trades in order to profit from the spread between buy and sell prices. This trading strategy is highly competitive in financial markets, necessitating constant optimization of algorithms and technological infrastructure to maintain a leading edge.

The rise of high-frequency trading (HFT) can be attributed not only to the proliferation of high-speed computers but also to a series of regulatory changes. In 1998, the U.S. Securities and Exchange Commission (SEC) introduced the “Alternative Trading System (ATS) Rule,” which created conditions for competition between electronic trading platforms and large exchanges. Two years later, exchanges began quoting in increments of nearly 1 cent instead of 1/16th of a dollar, resulting in a narrowing spread between bid and ask prices and compelling profit-seeking traders reliant on spreads to explore other trading strategies.

While high-frequency trading experienced rapid growth, professionals and regulatory bodies started paying attention and conducting studies on HFT. In 2005, the SEC introduced the “National Market System (NMS) Regulation,” requiring trade orders to be publicly displayed nationally rather than just within individual exchanges. Additionally, exchanges were required to adopt written rules prohibiting members from profiting through automated quotes across exchanges.

In April 2010, the SEC announced plans to discuss and consider implementing regulations on “high-frequency trading,” requiring HFT firms to report their identities and trading activities to the SEC. It was reported that the SEC could enforce a requirement for proprietary trading firms and large non-brokerage companies like hedge funds to use a single ID number for trades and provide information about their trading operations and market impact to the SEC. The previous year, the SEC had already suspended a prominent form of high-frequency trading known as “flash trading.”

In fact, the impact of high-frequency trading on the market has long been a subject of intense debate among investment banks. A report from the Federal Reserve Bank of Chicago indicated that approximately 70% of overall trading volume in the U.S. stock market is conducted through high-frequency trading, while the number of institutions engaged in HFT accounts for only 2%.

The Federal Reserve Bank of Chicago believes that although high-frequency trading contributes to market liquidity to some extent, program errors or human negligence can have disastrous effects on market trends. Presently, issues with high-frequency trading often stem from investors sending incorrect instructions to the machines. While the impact of such errors has been relatively limited so far, they have caused significant market fluctuations on multiple occasions.

High-frequency trading is a form of quantitative trading characterized by short holding periods, with position allocation determined by computerized quantitative models. The success of a high-frequency trading strategy is largely driven by its ability to process a large volume of data within a short period, which was not possible with manual trading in the past. Trading algorithms are typically kept strictly confidential by their owners, but many practical algorithms have proven their effectiveness in traditional trading. The competition in this field is not so much about developing groundbreaking algorithms but rather about executing algorithms faster. Below, we list some commonly used standard arbitrage strategies in high-frequency trading.

Market making involves placing a limit order to sell (or offer) above the current market price or a buy limit order (or bid) below the current price on a regular and continuous basis to capture the bid-ask spread. Automated Trading Desk, which was bought by Citigroup in July 2007, has been an active market maker, accounting for about 6% of total volume on both NASDAQ and the New York Stock Exchange.

Hidden within market data such as quotes and trading volumes, many pieces of information are often inadvertently concealed. By monitoring this data, computers have the potential to analyze and extract insights ahead of news reports, allowing for profitable opportunities. Since this information is publicly available and transparent, this strategy is entirely legal. It is a relatively traditional approach that involves monitoring the typical and atypical price changes and volume fluctuations of a large number of securities to generate appropriate buy and sell orders in anticipation of various events.

Another set of high-frequency trading strategies are strategies that exploit predictable temporary deviations from stable statistical relationships among securities. Statistical arbitrage at high frequencies is actively used in all liquid securities, including equities, bonds, futures, foreign exchange, etc. Such strategies may also involve classical arbitrage strategies, such as covered interest rate parity in the foreign exchange market, which gives a relationship between the prices of a domestic bond, a bond denominated in a foreign currency, the spot price of the currency, and the price of a forward contract on the currency. High-frequency trading allows similar arbitrages using models of greater complexity involving many more than four securities.

Company news in electronic text format is available from many sources including commercial providers like Bloomberg, public news websites, and Twitter feeds. Automated systems can identify company names, keywords and sometimes semantics to make news-based trades before human traders can process the news.

A separate, “naïve” class of high-frequency trading strategies relies exclusively on ultra-low latency direct market access technology. In these strategies, computer scientists rely on speed to gain minuscule advantages in arbitraging price discrepancies in some particular security trading simultaneously on disparate markets.

Another aspect of low latency strategy has been the switch from fiber optic to microwave and shortwave technology for long distance networking. The switch to microwave transmission was because microwaves traveling in air suffer a less than 1% speed reduction compared to light traveling in a vacuum, whereas with conventional fiber optics light travels over 30% slower.

High-frequency trading strategies may use properties derived from market data feeds to identify orders that are posted at sub-optimal prices. Such orders may offer a profit to their counterparties that high-frequency traders can try to obtain. Examples of these features include the age of an order or the sizes of displayed orders. Tracking important order properties may also allow trading strategies to have a more accurate prediction of the future price of a security.

Most algorithms are implemented using modern programming languages. Simple models may rely on basic univariate linear regression, while more complex ones can incorporate game theory, pattern recognition, and prediction algorithms. Neural networks and genetic algorithms have also been used to implement these models.

Given that high-frequency trading requires the execution of buy and sell orders at high speeds, computer manufacturers provide high-performance computer systems, including overclocked CPUs, to enhance performance.