The term “High-Frequency Trading” has taken on an air of mystery to people outside the world of electronic trading. For some, it may conjure imagery of shadowy forces secretly controlling the global stock markets. Others might have heard it’s possible to get rich without ever leaving their computer just by booting up trading software. The reality of HFT is, of course, less grandiose. While high frequency trading offers an excellent business opportunity for those with the right skill-set and resources, it is not the next step in progression for a novice trader. It requires a significant investment of time and capital as well as existing business relationships to have order flow executed. Anyone enticed by the allure of high-frequency trading and hungry for a piece of the pie should take a look at what it means to enter the high-frequency arena and whether it makes sense for their firm.
Demystifying High-Frequency Trading
High-Frequency Trading (HFT) is a form of algorithmic trading in which thousands of orders are executed every second using sophisticated computer programs. High-end hardware combined with skillful software engineering allows these applications to analyze a flood of live data across multiple exchanges in order to identify and capitalize on extremely short-lived market conditions. Because these fleeting trade opportunities only exist for milliseconds at a time, speed becomes paramount to success: the system with the shortest time to execution will generally be the most profitable. As a result, every part of a high-frequency system must be lightning-fast.
Minimum Requirements
Due to the need to maximize computing power and minimize end-to-end latency, the bulk of the upfront cost for establishing a high-frequency trading platform comes from building the necessary technological infrastructure, which includes:
- High-end, regularly updated computer hardware
- Co-located server space, rented for a premium within exchange data centers in order to be as physically close to exchange servers as possible
- Reliable, real-time market data feeds (with costly subscription fees)
- Trading algorithms that require a competent engineering team to implement and maintain
Finding the Right Talent
To build out high-frequency platforms, firms need to acquire expertise in a variety of areas. Professionals at HFT firms tend to be highly educated, often possessing advanced degrees in computer science, mathematics, physics or related fields. Additionally, many are poached from other relevant industries, such as telecommunications or data science, and bring valuable insight to the varied challenges of developing an electronic trading platform. Those with experience working directly with internal exchange architecture are much sought after as well.
Because of the technological requirements inherent in these systems, the backbone of a high-frequency operation will be its engineers. The most pressing concern for creating a team is to find software engineers skilled in enterprise-class programming languages (such as C++ or Java) who will develop the core application and trading algorithms. A constant development pipeline for creating new strategies and improving existing ones is necessary to avoid being left in the dust by competitors. Hardware engineers have grown increasingly valuable as many proprietary HFT platforms turn to customized hardware such as field-programmable gate arrays (FPGAs) to increase the efficiency and throughput of their applications. Last but not least, network engineers are a valuable asset. As mentioned previously, latency to the exchange plays a key factor in the success of a high-frequency system. Talented engineers who can optimize transmission speeds will prove crucial to achieving minimal delay between order and execution.
On top of the engineering needs, developing new trading strategies calls for deep familiarity with the workings of the financial industry and electronic markets. There are different approaches that HFT systems use to generate profit (for example, liquidity reimbursements and statistical arbitrage). Without a firm grasp of these mechanisms, an effective trading model will be nigh unattainable. This is where quantitative analysts (aka quants) come in. They use technology-driven mathematic and statistical analysis to develop models of financial markets which can be applied to the creation of algorithmic trading strategies. Quants understand the minute details of how markets function better than anyone; furthermore, most also possess advanced programming knowledge. The skills they bring to the table make them vital to any HFT development team.
Staying on Top
New entrants to high-frequency trading should be aware that building a trading system is only half the battle. Staying on the top of the industry will be a continuous project but is necessary to maintain a profitable market position.
“Innovate or die” holds especially true for high-frequency firms. The fierce competition inherent in high frequency trading has led to a perpetual arms race to develop the fastest, most cutting-edge trading systems modern technology allows. While this power struggle is largely hardware driven, data science is also an integral part of building profitable strategies. In the last decade or so, machine learning has become a tool for analyzing data sets greater in size and complexity than humans could possibly process on their own. Like the advent of nuclear weapons, the use of AI to refine algorithms has the potential to send the arms race into overdrive, making it more difficult (or even impossible) for exclusively human-developed strategies to keep pace. Holding a competitive position within the marketplace will only become more challenging as time goes on.
On the flip side, participation in this rapid advancement must be done with an ample degree of caution. When thousands of trades are occurring every second, failure can be fast and sudden. Errant behavior from a trading system can lead to catastrophic results before it is even possible to take action to address it. Knight Capital Group suffered the best known example of such a disaster when the forgetfulness of a single technician left outdated code running on a live server. The algorithm went haywire and caused a major disruption in stock prices on the New York Stock Exchange. When all was said and done, Knight Capital Group lost $440 million, and its market valuation collapsed. In order to mitigate these risks and avoid becoming the next KCG, HFT firms need to establish thorough testing procedures, including robust backtesting, scenario testing, and simulated live trading (known as paper-trading).
Aside from the many technological concerns already discussed, HFT firms must also keep apprised of changes both in the financial industry at-large and in government. There is a quagmire of regulatory requirements and associated reporting that is different for every country. Additionally, exchanges have their own sets of rules and requirements for market participants that are updated on a regular basis and must be followed for risk of not being allowed to trade at all. Tracking all of this may prove burdensome for many firms, but there is no way around it.
Facing the Challenge
Not every firm will be able to overcome the hurdles of developing a high frequency platform. Between mustering capital to build out the necessary infrastructure and assembling a team of talented engineers, the barrier for entry is already steep. While many HFT firms employ proprietary in-house software, third-party products can help ease some of the development burden by providing out-of-the-box functionality for essential components such as FIX routing, risk controls, position management, and market data providers. However, there is not much sense in seeking out prefabricated algorithms given that trading systems will have to target rare market opportunities in order to be successful. High frequency platforms should consequently be built around unique strategies, either developed wholly in-house or in tandem with a third-party vendor. Regardless of which approach an aspiring HFT firm decides to take, there is no escaping the fact that a wide breadth of both financial and technical expertise will be needed.
