Business, Legal & Accounting Glossary
HFT is a type of investment strategy whereby stocks are rapidly bought and sold by a computer algorithm and held for a very short period, usually seconds or milliseconds.
The advancement of technology over the last two decades has altered how markets operate. No longer are equity markets dominated by humans on an exchange floor conducting trades. Instead, many firms employ computer algorithms that receive electronic data, analyze it, and publish quotes and initiate trades. Firms that use computers to automate the trading process are referred to as algorithmic traders; HFTs are the subset of algorithmic traders that most rapidly turn over their stock positions. Today HFT makes up a significant portion of U.S. equities market activity, yet the academic analysis of its activity in the financial markets has been limited.
Automated trading, defined as electronic trading using algorithms at some stage in the trade process, has grown rapidly over the past decade and is still evolving. Commonly referred to as algorithmic trading or Algo trading, it can be divided into two main strands:
Typically, HFT firms generate earnings from doing a large number of small-size, small-profit trades. The small trade sizes, in part a consequence of operating with low latency (see below), implying that HFT firms take little risk per trade compared with traditional market-makers. The risk holding period is also very short, usually well under five seconds and frequently less than one second. As such, HFT requires a liquid underlying market.
The term market making refers to the strategy of quoting a simultaneous buy and sell limit order (quote) for a financial instrument in order to profit from the bid-ask spread. This can be either imposed by mandatory requirements set by market operators/regulators for entities covering that role (e.g. an official market maker such as the Designated Market Maker at the NYSE or Designated Sponsors at the Frankfurt Stock Exchange via the trading system XETRA) or voluntarily, i.e. without a determined obligation to quote. Several different terms are used to denote this kind of designated liquidity provision, e.g. market making with obligations, designated market making and registered market maker. Market makers frequently employ ―quote machines‖ which provide the respective electronic markets with their quotes. Quote machines are programs which generate, update and delete quotes according to a pre-set strategy. Due to the varying degree of sophistication among these programs, some of them employ techniques similar to HFTs, while others rely on the involvement of a human market maker. Since market making is a well known HFT strategy (Tradeworx 2010a).
Quantitative portfolio managers use quantitative models to form investment portfolios. Chincarini and Kim define quantitative (equity) portfolio management in the following way:
“The central, unifying element of quantitative equity portfolio management (QEPM) is the quantitative model that relates stock movements to other market data. Quantitative equity portfolio managers create such models to predict stock returns and volatility, and these predictions, in turn, form the basis for selecting stocks for the portfolio.” (Chincarini and Kim 2006)
In contrast to HFTs, QPMs frequently hold positions for extended periods of time, whereas HFTs tend to liquidate their positions rapidly and usually end trading days without a significant position (flat).
Compared to AT and HFT, QPM has a higher degree of human intervention. QPMs use algorithms to generate trading decisions based on statistical calculations and data analysis techniques. While QPMs automate the process of portfolio selection and the generation of trading signals, a human portfolio manager will usually validate the results of his quantitative model before transferring it to a (human or automated) trader for execution.
In fragmented markets, real-time investigation of different accessible order execution venues and of available order limits and quotes can improve execution results in agent and proprietary trading. Smart order routing (SOR) systems enable to access multiple liquidity pools to identify the best order routing destination and to optimize order execution (Ende 2010). They scan pre-defined markets in real-time to determine the best bid and offer limits or quotes for a specific order, thereby achieving the best price or other pre-defined execution benchmarks.
The smart order router selects the appropriate execution venue on a dynamic basis, i.e. real-time market data feeds are used by a rule framework. Such provisions support a dynamic allocation of the order to the execution venue offering the best conditions at the time of order entry including or excluding explicit transaction costs and/or other factors (e.g. the current technical latency of the venue). In order to achieve the best result in order execution on a real-time basis, i.e. price and explicit execution costs, two steps are required: first, at order arrival a routing system of an investment firm has to screen the respective execution venues for their order book situations, i.e. the execution price dimension. Second, the system has to incorporate a model that enables to calculate the total execution price of trades in different markets including applicable trading, clearing and settlement fees or even taxes, i.e. the explicit costs dimension (Domowitz 2002).
The Scope of Algorithmic Trading Strategies
HFT is mostly defined as a subset of AT strategies. However, not all algorithmic strategies are necessarily high frequent. Most non-HFT algorithmic strategies aim at minimizing the market impact of (large) orders. They slice the order into several smaller child orders and spread these child orders out across time (and/or venues) according to a pre-set benchmark. The following subsections describe some of the more common non-HFT algorithmic Strategies
The classification into four generations is based on Almgren (2009) and includes information from Johnson (2010). First-generation algorithms focus solely on benchmarks that are based on market generated data (e.g. VWAP) and are independent from the actual order and the order book situation at order arrival, while the second generation tries to define the benchmark based on the individual order and to handle the trade-off between market impact and timing risk. Third generation algorithms are furthermore able to adapt to their own performance during executions. A fourth-generation – that is not included in the Almgren (2009) classification – consists of so-called newsreader algorithms.
Participation Rate Algorithms
Participation rate algorithms are relatively simple. They are geared to participate in the market up to a predefined volume. Such an algorithm could, for example, try to participate by trading 5% of the volume in the target instrument(s) until it has built or liquidated a target position. Since these algorithms target traded volume, they reflect the current market volume in their orders. Variants of these algorithms add execution periods during which orders are submitted to the market or maximum volumes or prices.
Furthermore, randomized participation rates are used to make the algorithm harder to detect for other market participants.
Time Weighted Average Price (TWAP) Algorithms
TWAP algorithms divide a large order into slices that are sent to the market in equally distributed time intervals. Before the execution begins, the size of the slices, as well as the execution period,s defined. For example, the algorithm could be set to buy 12,000 shares within one hour in blocks of 2,000 shares, resulting in 6 orders for 2,000 shares which are sent to the market every 10 minutes. TWAP algorithms can vary their order sizes and time intervals to prevent detection by other market participants.
Volume Weighted Average Price (VWAP) Algorithms
VWAP algorithms try to match or beat the volume-weighted average price (their benchmark) over a specified period of time. VWAP can be calculated applying the following formula for n trades, each with an execution price pn and size vn (Johnson 2010):
VWAP = (Overall Turnover) / (Total Volume) = Summation vn pn / Summation vn
Since trades are being weighted according to their size, large trades have a greater impact on the VWAP than small ones. VWAP algorithms are based on historical volume profiles of the respective equity in the relevant market to estimate the intraday/target period volume patterns.
The most prominent second-generation algorithms try to minimize implementation shortfall. The current price/midpoint at the time of arrival of an order serves as a benchmark, which shall be met or outperformed (order based benchmark). Implementation shortfall algorithms try to minimize the market impact of a large order taking into account potential negative price movements during the execution process (timing risk). To hedge against an adverse price trend, these algorithms predetermine an execution plan based on historical data, and split an order into as many as necessary but as few as possible suborders. In contrast to TWAP or VWAP, these orders will be scattered over a period which is just long enough to dampen the market impact of the overall order (Johnson 2010).
Adaptive algorithms form the third generation in Almgren‘s classification (Almgren 2009). These algorithms follow a more sophisticated approach than the implementation of shortfall algorithms. Instead of determining a pre-set schedule, these algorithms re-evaluate and adapt their execution schedule during the execution period, making them adaptive to changing market conditions and reflecting gains/losses in the execution period by a more/less aggressive execution schedule.
Investors have been relying on the news to make their investment decisions ever since the first stock market opened its gates. Since then, traders who possess valuable information have been using it to generate profits. However, there is a limit to the quantity of data a human trader can analyze, and maybe even more important, the human nature of an investor/trader limits the speed with which he/she can read incoming news. This has led to the development of newsreader algorithms. These automated newsreaders employ statistical methods as well as text-mining techniques to discern the likely impact of news announcements on the market. Newsreader algorithms rely on high-speed market data. Exchanges and news agencies have developed low latency news feeds, which provide algorithmic traders with electronically processable news.
While consolidated information on the major players in HFT is still scarce, the community of market participants leveraging HFT technologies to implement their trading strategies is highly diverse. Its members range from broker-dealer operated proprietary trading firms and broker-dealer market-making operations to specialized HFT boutiques to quantitative hedge funds leveraging HFT technology in order to increase the profits from their investment and trading strategies (see Easthope and Lee 2009). There is (i) a multitude of different institutions with different business models that use HFT and (ii) there are many hybrid forms, e.g. broker-dealers which run their proprietary trading books applying HFT techniques. Therefore, in the assessment of HFT, it is very important to take a functional rather than an institutional perspective. In order to achieve a level playing field, all institutions that apply HFT based trading strategies have to be taken into consideration independent of whether HFT is their core or an add-on technology to implement trading strategies.
While the universe of HFT strategies is to diverse and opaque to name them all, some of these strategies are well known and not necessarily new to the markets. The notion of HFT often relates to traditional trading strategies that use the possibilities provided by state-of-the-art IT. HFT is a means to employ specific trading strategies rather than a trading strategy in itself. Therefore, instead of trying to assess HFT as such, it is necessary to have a close look at the individual strategies that use HFT technologies
High Frequency Based Trading Strategies
Electronic Liquidity Provision
One of the most common HFT strategies is to act as a liquidity provider. While many HFTs provide the market with liquidity like registered market makers, they frequently do not face formal obligations to quote in the markets in which they are active. HFT liquidity providers have two basic sources of revenues: (i) They provide markets with liquidity and earn the spread between the bid and ask limits and (ii) trading venues incentivise these liquidity provides by granting rebates or reduced transaction fees in order to increase market quality and attractiveness.
A HFT strategy, which closely resembles its traditional counterpart, i.e. market-making, is spread capturing. These liquidity providers profit from the spread between the bid and ask prices by continuously buying and selling securities (ASIC 2010a). With each trade, these liquidity providers reap the spread between the (higher) price at which market participants can buy securities and the (lower) one at which they can sell securities
Rebate Driven Strategies
Other liquidity provision strategies are built around particular incentive schemes of some markets. In order to attract liquidity providers and react to increasing competition among markets, some trading venues have adopted asymmetric pricing: members removing liquidity from the market (taker; aggressive trading) are charged a higher fee while traders who submit liquidity to the market (maker; passive trading) are charged a lower fee or are even provided a rebate. An asymmetric fee structure is supposed to incentivize liquidity provision. A market operator‘s rationale for applying maker-taker pricing is given by the following: traders supplying liquidity on both sides (buy and sell) of the order book earn their profits from the market spread. Fee reductions or even rebates for makers shall stimulate a market‘s liquidity by firstly attracting more traders to post passive order flow in the form of limit orders. Secondly, those traders submitting limit orders shall be incentivized and enabled to quote more aggressively, thus narrowing the spread. The respective loss of profits from doing so is supposed to be compensated by a rebate. If this holds true, those markets appear favourable over their rivals and market orders are attracted enhancing the probability for the makers to have their orders executed (Lutat 2010).
Opportunities to conduct arbitrage strategies frequently exist only for very brief periods (fractions of a second). Since computers are able to scan the markets for such short-lived possibilities, arbitrage has become a major strategy applied by HFTs. These HFTs conduct arbitrage in the same way as their traditional counterparts; they leverage state of the art technology to profit from small and short-lived discrepancies between securities. The following types of arbitrage are not limited to HFT, but are conducted by non-automated market participants as well. Since arbitrageurs react on existing inefficiencies, they are mainly takers of liquidity.
Market Neutral Arbitrage
This form of statistical arbitrage aims to be ―market neutral‖. Arbitrageurs try to hold instruments while simultaneously shorting other instruments. Since the instruments are closely correlated, gains and losses due to movements of the general market will (mostly) offset each other. However, in order to gain from this strategy, arbitrageurs sell an instrument which they deem to have a relatively lower intrinsic value, while simultaneously buying an instrument, which reacts very similar (ideally identical) to changes in the market environment and which they deem to have a relatively higher intrinsic value. If the respective valuation of these instruments ―normalizes‖ into the expected direction, the arbitrageur liquidates its market neutral position. Gains from this strategy result from the difference between the individual valuation of the assets at the time the position is opened and their ―normalized‖ prices at the time the position is liquidated. Since this strategy offers protection against market movements, it is highly attractive for HFTs and traditional arbitrageurs alike. (Aldridge 2010)
Cross Asset, CrossMarket & Exchange Traded Fund (ETF) Arbitrage
An established arbitrage strategy is to trade instruments across markets or to trade-related instruments and to profit from pricing inefficiencies across markets: if an asset shows differing prices across marketplaces, arbitrageurs generate profits by selling the asset on the market where it is valued higher and simultaneously buying it on another market where it is valued lower.
Cross market arbitrage strategies have profited from the increased market fragmentation in Europe as described in section two. A higher number of markets increases the probability that an instrument has different prices across these markets. Similarly, arbitrageurs can profit from inefficiencies across assets: if, e.g. an option is priced too high relative to its underlying; arbitrageurs can earn profits by selling the option and simultaneously buying the underlying. In a similar way, ETF arbitrageurs trade ETFs against their underlying and profit from respective pricing inefficiencies. Since such inefficiencies exist only shortly on modern securities markets, HFTs leverage their speed advantage to trade against them
Another category of HFT strategies is liquidity detection. These HFTs try to discern the patterns other market participants leave in the markets and adjust their actions accordingly. Liquidity detectors focus their attention on large orders and employ various strategies to detect sliced orders, hidden orders, orders being submitted by execution algorithms or to gain further information about electronic limit order books (ASIC 2010a). Liquidity detectors gathering information about algorithmic traders are frequently referred to as ―sniffing out‖ other algorithms. Other detectors ―ping‖ or ―snipe‖ in order books or dark pools to retrieve information from them (see e.g. ASIC 2010a).
Another possible way to use HFT technology would be a high-speed version of the ―quote matching‖ strategy described by Harris (2003). Using this strategy, a trader who has detected a large order within the order book places his own order ahead of the large order. If he has detected for example a large buy order, he places his own buy order at a slightly higher limit. Should prices now move upwards, he profits from the rise. However, should prices fall, the large order resting in the book serves as an option/hedge against which the trader can sell his own shares, thereby limiting his possible losses as long as the large limit order rests within the book.
Other High-Frequency Trading Strategies
Some market participants accuse HFTs of conducting a form of arbitrage which is purely based on their faster access to market data. This modern form of arbitrage, where HFTs are said to be able to see (and interpret) new market information before many market participants even receive it, is frequently referred to as latency arbitrage. These latency arbitrageurs leverage direct data feeds and co-located infrastructure to minimize their reaction times. Especially in the U.S., where many market participants rely on the ―national best bid and offer‖ (NBBO) , latency arbitrageurs are said to be able to profit from their speed advantage in comparison to the NBBO (see e.g. Gaffen 2009).
Short-Term Momentum Strategies
Market participants leveraging HFT technologies to conduct short-term momentum strategies are a modern equivalent to classical day traders. In contrast to many other HFT based strategies they are neither focused on providing the market with liquidity, nor are they targeting market inefficiencies. They usually trade aggressively (taking liquidity) and aim at earning profits from market movements/trends. Their trading decisions can be based on events influencing securities markets and/or the movements of the markets themselves. Momentum based trading strategies are not new and have been implemented by traditional traders for a long time.
To help you cite our definitions in your bibliography, here is the proper citation layout for the three major formatting styles, with all of the relevant information filled in.
Definitions for High-Frequency Trader are sourced/syndicated and enhanced from:
This glossary post was last updated: 14th April, 2020 | 2 Views.