The current discourse on high-frequency trading is often challenged by a distortion of definitions. Journalists, politicians and industry analysts bend or stretch definitions to meet their various (and often conflicting) objectives. For example, flash orders and high-frequency trading have been improperly used as equivalent terms. Front-running has been invoked when "liquidity detection" would be more accurate. While there is room for a legitimate debate over the scope, size and impact of high-frequency trading, the industry must first agree to terms. Below, TABB Group outlines a few principles to which it adheres when discussing this controversial subject:
HFT refers to fully automated trading strategies (in equities, derivatives or currencies) that seek to benefit from market liquidity imbalances or other short-term pricing inefficiencies. These opportunities could last from milliseconds to minutes and possibly hours. While these strategies can be employed overnight, the majority of HFT strategies attempt to be market-neutral or closed out by the end of each day.
The kinds of strategies that fall under HFT include electronic market making, liquidity detection, cross-asset arbitrage, short-term statistical arbitrage and volatility arbitrage. The most prevalent equity HFT strategy is electronic market making, in which firms attempt to profit from intraday imbalances in the supply and demand for liquidity. Not all market making is high-frequency (though almost all of it is), and not all high-frequency trading is market making, but market-making strategies profit by intelligently managing the risk caused by inconsistencies between buyers and sellers.
Perhaps the most controversial and least understood aspect of high-frequency trading falls under the category of liquidity detection. While classic market makers attempt to capture spread by aggressively quoting at the bid and the ask of a number of stocks, a liquidity detector uses techniques to sniff out large orders of blocks being sliced and diced (usually by an algorithm) that a high-frequency trader believes it can outsmart.
Who Does It?
Although HFT makes up a large portion of total trading activity, a relatively small number of firms are responsible for its volume. Three types of firms build their strategies around HFT: proprietary trading firms (virtual market makers), the largest hedge funds and investment banks' proprietary trading desks. While each of these institutions has a unique position in the industry, their common ground is their mandate to achieve uncorrelated and high returns.
Approximately one-half of liquidity provisioning these days comes from traditional market makers or large broker-dealers. The remainder originates from low-profile (though this is now changing) high-frequency trading firms -- the proprietary (prop) trading shops -- that few other than the industry intimates have ever heard of. Prop shops have been around for many years, earning their profits by risking their own capital. They originated either from groups formerly within broker-dealers or independent firms that have the knowledge, skills and technology to fully automate the trading process; or from screen-based day-trading shops that began automating their strategies in the late 1990s/early 2000s. These prop shops virtually automated the market-making function by leveraging inexpensive computing cycles, low-latency infrastructures and fully automated trading strategies.
Most HFT prop shops choose to keep their identities and intentions secretive, operating under the radar in the hope of improving their chance to profit. Through a thorough examination of Web sites and other public information, TABB Group has found that while the vast majority of these firms trade U.S. equities, the firms are quick to apply their strategies to the entire array of asset classes (see chart).
Investment banks have always traded for their own accounts. Their prop desks typically operate from a distinct legal entity -- separate from the entity that handles customer orders -- within the investment bank; the bank risks its own capital by deploying trading strategies designed to maximize profit. Two divisions within investment banks that deploy HFT are automated market making and proprietary desks. Market makers are registered with the SEC, using traditional trading strategies to facilitate liquidity in the market. Prop desks implement a variety of arbitrage strategies, some of which are high-frequency (though certainly not exclusively high-frequency).
For the most part, high-frequency hedge funds engage in short-term trading opportunities rather than bona fide liquidity-based strategies. While the umbrella term statistical arbitrage is frequently applied to strategies with extremely high volumes, there is plenty of ambiguity in this term. It is also true that the majority of funds engaged in statistical arbitrage are not high-frequency by today's standards. However, over the past 18 months the line between high-turnover strategies and HFT has blurred as hedge funds shorten their time horizons in the face of unexpected market events.
As a result, transaction costs are becoming even more paramount to this sophisticated community. The rationale is that as time horizons shorten, capacity constraints increase and transaction costs become a bigger piece of the pie. High-frequency hedge funds may be layering these liquidity strategies on top of their other strategies so that transaction costs are additive rather than negative.
How Big Is It?
The only art more forgivable than economic forecasting is estimating the market size of an industry that will never reveal its true number. Nonetheless, TABB Group estimates that high-frequency trading accounts for 61 percent of U.S. equity share volume (remember to double-count average daily shares!) and generates $8 billion per year in trading profits.
The methodology begins with an analysis of institutional equity trading volume that we have been collecting since 2006 from 115 U.S.-based equity head traders, including equity assets under management, average daily volume and the percentage of shares executed in blocks. We extrapolate that data to the broader institutional landscape. Retail trade numbers and data from the government are used to determine retail flow. Data from NYSE and Nasdaq and historical market making volumes enhances our picture of current electronic market-making volumes. Last but not least, we discussed our methodology and trading profit calculations (.0024/share) with several HFT hedge funds, independent high-frequency traders and registered market makers.
Is It Good for the Market?
This is the wrong question. The right questions are whether the current market structure can be improved, and what the role of HFT should be in any revised market structure. But that is a scary question because outside of consulting (ahem), IT and perhaps the end investors, there is little for the industry to gain out of major changes to market structure.
The market structure changes and technological advances over the last decade that have made it possible for virtual market makers to supplant the traditional players are viewed as primarily positive for the market. Very few participants or observers suggest that we should roll back the clock on decimalization and exchange competition. Participants feel today's market structure is orderly despite its complexity, and that it does a very good job of encouraging price discovery (see chart).
High-frequency equity trading is the lovechild between 12 years of SEC rulemaking and advances in trading technology. The combination of these two trends has been necessary and sufficient to unleash an array of new trading strategies. The continued success of these strategies has exchanges and ECNs, brokers and clearinghouses, and market data providers and technology vendors launching new business models and offerings to support high-frequency traders or to help others adapt to this new environment. Imagining a U.S. equity market structure without high-frequency traders is like trying to remove the c from E=mc2.
Adam Sussman is director of research for TABB Group. Previously he served as a senior product manager at Ameritrade, where he was responsible for order management systems, routing and next-generation trading tools focused on the equities and options markets.