2019 Curriculum CFA Program Level II Portfolio Management and Wealth Planning
Algorithmic Trading and High-Frequency TradingView the full reading
Available to members
It is estimated that 75% of US stock trades are not placed by humans but by computer algorithms. This figure has been expanding over time and is expected to continue to do so. More trading is done by machines than humans because the human brain cannot process the volumes of information needed to make trading decisions and place trades before a competitor does. Algorithms can process millions of pieces of data per second, make sub-millisecond decisions, and take autonomous actions.
A trading algorithm may be as straightforward as an execution algorithm that is programmed to intelligently slice up large trades on behalf of a buy-side firm (such as a pension fund or mutual fund) to minimize market impact. But an algorithm can get as complex as a self-learning, high-frequency algorithm that makes decisions on what, when, and how to trade and executes these trades itself, without any human input.
It is not just equities that are traded by algorithms; the same algorithmic trading trend is evident in other electronically traded asset classes: futures, foreign exchange (FX), bonds, energy, and so on. In all of these asset classes, algorithms are autonomously managing more and more of the trading decisions. And this trend is occurring in all trading markets around the world. There is, in fact, a high- frequency algorithmic war raging:
Algorithms compete to find the best opportunities and execute on them first. This has been a concern to some parties who are worried that certain market participants have an “unfair advantage.” But humans are still needed as the creators of algorithms and arbiters of good sense. It has not yet become possible to digitize the instincts of a really good trader!Algorithms have a life cycle: from research to implementation to testing to tuning. Sometimes algorithms go wrong, which can be extremely costly. There is, therefore, increased interest in using compliance algorithms to monitor trading algorithms, with a view to detecting aberrant behavior.
The candidate should be able to:
- define algorithmic trading;
- distinguish between execution algorithms and high-frequency trading algorithms;
- describe types of execution algorithms and high-frequency trading algorithms;
- describe market fragmentation and its effects on how trades are placed;
- describe the use of technology in risk management and regulatory oversight;
- describe issues and concerns related to the impact of algorithmic and high-frequency trading on securities markets.
Algorithmic and high-frequency trading are important factors in today’s markets. Just like electronic terminals replaced open outcry (trading by shouting and waving bits of paper in the trading pits of stock exchanges), so algorithms are replacing the humans that operated the electronic trading terminals in various forms of trade execution. Key points to remember regarding algorithmic trading include the following:
- There are two main types of algorithms: execution algorithms, which minimize the market impact of large orders, and high-frequency algorithms, which constantly monitor real-time market data and look for patterns to trade on.
- Algorithms can adapt to market fragmentation by incorporating liquidity aggregation and intelligent smart order routing capabilities.
- Algorithms can be used for real-time pricing of instruments.
- Low latency is important and latency at each layer of the end-to-end latency equation must be considered: the physical connections to the market, the market data feeds, the algorithmic engine, and the order execution feed to a trading venue.
- The life cycle of an algorithm includes alpha discovery to find new patterns, algorithm implementation, back testing, production, and tuning.
- Algorithms are used in many asset classes, including equities, futures, foreign exchange, bonds, and energy. Algorithms will likely be developed to exploit additional areas as new types of assets migrate to electronic trading.
- Surveillance algorithms can be used to spot potential market abuse and compliance breaches.
- The broad market impact of algorithmic trading is largely positive. Research shows that HFT has led to tighter bid–ask spreads, lower transaction costs, increases in liquidity, and improved pricing efficiency.
- The primary concerns regarding HFT are the potential for HFT to accentuate and accelerate market movements: the risk posed by an out-of-control algorithm, the ability of a trader to manipulate the market through spoofing or quote stuffing, the increased complexity of regulatory oversight, and the impact of unequal access to information.