High-frequency trading (HFT) is a method of trading that uses powerful computer programs to transact many orders in fractions of a second. It uses complex algorithms to analyze multiple markets and execute orders based on market conditions. Typically, the traders with the fastest execution speeds are more profitable than traders with slower execution speeds. In addition to the high speed of orders, HFT is also characterized by high turnover rates and order-to-trade ratios.
HFT is not just one strategy; it is a conglomeration of several different strategy types. These strategy types fall into different categories like market making, arbitrage, momentum, news-driven HFT, and many more. Let’s look at the categories listed here to help understand how broad the spectrum of HFT trading types is.
High-Frequency Trading Categories
- “Market making” can have two different meanings. There is the official capacity of market makers, also formerly known as specialists, who are firms or individuals that actively quote two-sided markets in a security. There are also unofficial market makers who fall into the HFT category by providing simultaneous quotes on both the bid and offer. These HFTs use complex algorithms that measure buying and selling pressures and modify the bids and offers rapidly, looking to earn small fractions of a penny repeatedly. They also may look to collect trading rebates through the maker-taker rebate model. This is a fully electronic model, with no human intervention, due to the rapid adjustment of orders on a sub-second time frame. Both types of market makers need to consider the high costs of requiring raw market data from the exchanges, before acting in this capacity.
- Arbitrage utilizes trading algorithms to identify price discrepancies in the markets. Arbitrage HFT can be divided into statistical, event-driven, and index arbitrage subcategories, but the common thread is the use of algorithms to capitalize on securities being out of balance in different exchange locations. Simply put, security trading in multiple locations can occasionally trade at different values. Arbitrage algorithms attempt to buy undervalued securities and simultaneously sell the same securities at another location where it is overvalued, locking in a small profit. Repeating this process helps bring the valuations into line, providing a valuable market function while simultaneously netting the HFT a profit. Arbitrage, however, can be a crowded space with many participants all trying to capitalize on the same price discrepancy, so it can be a challenge to capitalize regularly on arbitrage opportunities before the prices level out.
- News-driven HFT is based on the concept of scanning financial news and other breaking news sources for key headlines and content that may trigger moves in the markets. Once a scanning algorithm identifies breaking news and the potential effect on the market direction, it will sweep up orders in the hopes of being ahead of the market reaction. The biggest challenges with news driven strategies are receiving data quickly and reacting correctly, as there can be different interpretations of the same “news”. Having a slow inbound connection or latency outbound to the exchanges will have an impact on your ability to profit regularly from a news event as well.
- Momentum HFT is like news-driven HFT in terms of being ahead of the crowd, but these algorithms are looking for key price action and technical setups in the markets in conjunction with order flow components. Once the algorithm identifies these setups, the algorithm will sweep up the resting orders in the hope that a momentum move will follow in the HFT’s direction. Once the move has been triggered, the HFT will unwind the position. An example would be a stock that meets a major resistance price level with little depth of market showing in the lit order book. A momentum HFT algorithm may attempt to buy up many of the resting orders, triggering a price breakout to the upside past the major resistance level, which will then be followed by other momentum algorithms and eventually retail traders. Although momentum driven opportunities appear to be a great alpha generator, they also present programing challenges. Correctly coding signals that convert directly to action in the markets is difficult. There can be many false signals using this strategy and that could be costly.
This isn’t even a comprehensive list of all the different HFT strategies out there, but all of them have one thing in common: they rely on speed. Most people think of speed in electronic trading as computing power, but that can be made into a level playing field. HFT traders have access to the same computers and computing power. The real advantage comes from the speed of light.
“Tick-To-Trade” Latency and Using Colocation for Your Benefit
Data transmission as we know it is limited to the speed of light. The farther away from something you are, the slower it will get there, even at the speed of light. This is where HFT uses colocation to their benefit. The closer you are to the exchange computers that control the execution and dissemination of data, the faster your HFT algorithms can get data, analyze it, and execute trades.
This is called “tick-to trade” latency. Put another way, it is the time interval between receiving a market tick showing opportunity to your algorithm and sending the buy/sell order.
Major Exchange Colocation Center Locations and the Effects on High-Frequency Trading
As the graphic shows, all the major exchange colocation centers are in the New York/New Jersey area. The closer the HFT computers are to these locations, the faster their access is to data. A computer in Chicago, even at nearly speed-of-light data transmission over fiber optic cable, is at a distinct disadvantage to an HFT computer located in New York or New Jersey — even the computing power is not equal. The HFT space is so competitive, there is even competition to get the HFT servers located as close to the necessary data sources as possible.
Moreover, there are multiple locations, so if the HFT strategies being deployed trade on multiple exchanges, multiple servers need to be at multiple colocation facilities and even then, the method of communication between the facilities comes into play. Some people think the fastest way to transmit data between colocation facilities is via fiber optics when microwave radio transmission is faster.
For example, transmitting data from two facilities 200 km apart via microwave is 1064 microseconds and via fiber is 1594 microseconds. The 530 microseconds make a substantial difference to HFT traders. Lime Execution’s offering is built to support HFT firms with their low-latency infrastructure.
Final Thoughts
So, in the end, speed matters, and the HFT traders and firms that could optimize data acquisition and utilize speed the best could have a distinct advantage over traders and firms that lose that race. Colocation can be the key for many strategies, not just HFT algorithms. Firms or traders that are not strategically colocated with the exchange servers are disadvantaged in the complex world of quantitative and high-frequency trading.
