• —What do experienced bond traders and typists have in common?
• —Computers ate their lunch.

High‐frequency trading (HFT) is another fintech innovation capitalizing on plunging costs of technology. When programmed correctly, HFT software has built‐in advantages over manual trading. Computers seldom become ill, are hardly emotional, and make, in short, superior cool‐headed traders who stick to the script and don't panic.

Of course, some recent research purports that some people, especially those attuned to their intuitive or biological responses, can outperform machines. Well, good for those few! By and large, however, human traders tend to be a superstitious, irrational lot prone to, well, human behavior, and no match for their steely automated trading brethren.

Perhaps one of the largest advantages of machines, however, is not that they can contain their nonexistent feelings, but in their information processing power. Humans have a finite ability to process data. We may, possibly, be able to stare at some 16 screens all at once, but our eyes can still only process 24 distinct visual frames per second. Should the information update faster than that, we simply miss it. Computers, on the other hand, can process unlimited volumes of data at the speed of light.

Even more importantly, following just a few news sources and several price charts in today's interconnected continuously arbitraged markets is simply not enough. News leaks out into the markets in chaotic and often unforeseeable ways. Processing the entire realm of information, including quotes for some 6,000+ stocks, hundreds of thousands of options, interest rates, futures, and social media is what separates today's successful traders from the not‐so‐successful ones. And machines simply do it better. No extreme human physiology allows us to simultaneously read in‐depth news and analyses on even 100 financial instruments.

The HFT, on the other hand, are generally well‐equipped to process reams of information on the fly. Still, as discussed in Chapter 4, all HFTs do not fit in the same mold. Some are market makers, using predominantly limit orders, passively waiting for the market‐order‐armed liquidity takers to arrive. Others are aggressively pursuing the best price available at a given point in time. Both categories, passive HFT and aggressive HFT, have their parallels in the world of human traders: passive HFTs are automated versions of their human market‐maker predecessors, and aggressive HFTs are modeled on former prop traders and day traders.

Passive HFTs, described in Chapter 4, are a set of HFT strategies mostly placing limit orders. As such, passive HFTs end up buffering market liquidity, and making markets. Prominent passive HFT firms include Virtu and Knight Capital Group. The market makers, whether human or robotic, follow the same basic principles. Market‐making strategies consist of often‐simultaneous placement of limit orders on both the buy and the sell side of the limit order book, shown in Figure 5.1. The simple two‐sided quotation works great when markets are quiet or range‐bound. However, when the markets move rapidly in one direction or another, market makers risk severe losses.

The market makers' risks are largely a result of their potential information asymmetry. For example, suppose that the latest trade price of IBM stock is $155.76, and a market maker is quoting $155.74/$155.77, as shown in Figure 5.2. Suppose further that another trader with better information strongly believes that the IBM market is about to fall, say, to $155.00 in the next 20 minutes. To capitalize on his fleeting informational advantage, the second trader places the market order to sell IBM stock. The trading venue matches the market sell order with the market maker's limit buy order at $155.74, leaving the market maker holding a long position in IBM in a rapidly falling market. Fast‐forward 20 minutes, and the market maker is in the hole at $0.74 for every share he bought. The phenomenon is known as adverse selection whereby the uninformed market makers and other traders are “picked off” by better‐informed market participants.

Is being better‐informed illegal? Of course not. Some superior information costs a pretty penny, and substantially reduces the informed traders' gains from trading, when netted out at the end of the day. What is a market maker to do? The market maker needs to (1) hedge his exposure, and (2) become better informed. Hedging exposure is always costly. For instance, the market maker may decide to purchase options or other derivatives to hedge his exposure to the underlying. A put option will do the trick, but at an upfront premium.

Another route is to obtain better information. One way of gearing up on the information frontier is segmenting traders into better‐informed and worse‐informed, and essentially front‐running better‐informed traders using prehedging or anticipatory hedging, as discussed in Chapter 4. The strategy works mostly at a broker‐dealer's market maker, since order flow on exchanges and other venues outside of a broker‐dealer are largely anonymous. Another, more ethical route comprises investing into premium data that is indicative of the market's near‐term direction, shrinking the information barrier and preempting sharp moves. Once again, computerized market makers win the bots‐versus‐humans debate as information is king, and the ability to process vast amounts of information simultaneously is cash.

Bots using market orders that trade using intricate strategies of professional traders are collectively known as aggressive HFT, as opposed to passive market makers. Unlike passive HFTs, aggressive HFTs tend to use market orders to capitalize on fleeting information at their fingertips. In the industry, the return advantage from fleeting information is often referred to as rapidly decaying alpha.

What kinds of inferences do aggressive HFTs deploy? To put it simply, all kinds. The most successful aggressive HFTs, like QuantLab and others, use a multitude of information sources to create an informational haystack, from which big data‐driven inferences are extracted about the prospective market movements.¹

Not everyone agrees with the premise that the usually‐private HFTs are a success. Recently, some articles have declared HFT dead. One author even suggested that “poor” HFTs should be “pitied” as their strategies have been wiped out by volatility and dog‐eat‐dog competition for faster, better, ever‐more expensive technology undermining all profits. According to the research of AbleMarkets.com, discussed here, however, nothing could be farther from the truth: aggressive HFTs quietly, but significantly, prosper, and more so in the currently volatile market conditions.

Just how profitable are aggressive HFTs? It's impossible to know the full range of profitability of all the HFT firms. However, just a set of aggressive HFTs that hold positions for one minute on average can achieve a Sharpe ratio of 39 (that's right, thirty‐nine) trading all 500 stocks in the S&P 500. Aggressive HFT in all of the 500 stocks produces nearly 35 percent return per year without any additional leverage, generating almost no volatility in the performance, regardless of market conditions.

Figure 5.3 illustrates the performance of selected aggressive HFT algorithms detected by the AbleMarkets.com Aggressive HFT Index. The index estimates the participation of aggressive HFT by volume in real time. The profitability of aggressive HFTs is computed as follows: Whenever the proportion of volume estimated to be initiated by aggressive HFT sellers exceeds the proportion of volume the index attributes to aggressive HFT buyers by at least 10 percent, a hypothetical sell order is generated and the position is “held” for the following 1 minute, at which point the position is considered to be closed at the prevailing market price and the gain or loss of the trade is recorded. Similarly, whenever the AbleMarkets Aggressive HFT Index reports the proportion of aggressive HFT buyers exceeding that of aggressive HFT sellers by 10 percent or more, a paper buy order is generated, followed by an offsetting sell order 1 minute later.

The resulting trading strategy illustrates just how good aggressive HFTs are at their game. While highly sensitive to transaction costs, a Sharpe ratio as high as 39 makes a compelling case for investors to invest in super‐fast processing power in order to maximize the net gain. The strategy may also be used to optimize the execution of other strategies with large positions in individual securities: deciding when and how to send in slices of large orders to exchanges, as discussed in the following section. Briefly, when the percentage of aggressive HFT buyers is higher than that of aggressive HFT sellers, aggressive HFTs wipe out liquidity in offers, but create a surplus of bids—it is a good time to withhold buy orders and place sell orders instead.

Should we be surprised by such performance results of HFTs? Not really—the performance is perfectly in line with HFT giants such as Virtu Financial, who have not experienced a single losing week in a few years. In other words, reports of premature death of HFT appear to be, well, premature.

Latest research shows that stocks with higher participation of aggressive HFT experience higher intraday volatility. Comparing intraday volatility, as measured by the difference between the daily high and low and normalized by the daily closing price, with aggressive HFT participation captured by AbleMarkets Aggressive HFT Index shows that every 1 percent increase in aggressive HFT participation by volume drives up same‐day volatility by 2 percent on average! As discussed earlier in this chapter, aggressive HFTs comprise a set of trading strategies that predominantly use market orders (as opposed to limit orders) to execute their trading decisions. As such, aggressive HFT strategies are usually characterized by short profitability windows and strive to capture gains as soon as possible with the most immediate order execution. How much does aggressive HFT participation change from one day to the next? According to AbleMarkets research, the answer is 80 percent of the time (across all the S&P 500 stocks), aggressive HFT increases or decreases by at most 3 percent from one day to the next. However, outliers exist, and aggressive HFT may spring up in some previously untapped names. For example, on February 26, 2015, aggressive HFT participation in Graham Holdings Company (NYSE: GHC) spiked up to 66 percent by volume from just 34 percent observed on the previous business day only to drop back down to 34 percent on the following trading day, February 27, 2015.

On days when the participation of aggressive HFTs across all of the S&P 500 stocks sparked up by 3 percent or more, observed intraday volatility on average was 2.4 percent with a standard deviation of 2.0 percent during the first six months of 2015. On the other hand, on days when the participation of aggressive HFTs across all the S&P 500 stocks declined by 3 percent or more from the previous trading day, observed intraday volatility on average was 1.5 percent with a standard deviation of 0.8 percent. For a baseline comparison across all days and across the S&P 500, observed intraday volatility was 1.7 percent on average with a standard deviation of 1.1 percent.

The aggressive HFT is “sticky.” Stocks with high aggressive HFT retain high aggressive HFT participation as HFT developers ramp up and down their algorithms slowly over time. In fact, the previous day's aggressive HFT participation is a great predictor of the next trading day's aggressive HFT. The analysis of the S&P 500 shows that a daily value of AbleMarkets Aggressive HFT Index explains 86 percent of variation in the aggressive HFT participation (measured by AbleMarkets Aggressive HFT Index) on the following trading day, as measured by a statistical metric known as Adjusted R‐squared.

How do these findings translate into the prediction of volatility? Both day‐to‐day changes and absolute values of the AbleMarkets Aggressive HFT Index are predictive of the next day's volatility. AbleMarkets Aggressive HFT Index helps to make the detection of aggressive HFT and the resulting volatility prediction easier and more manageable.

IMPLICATIONS OF AGGRESSIVE HFT

As a natural consequence of aggressive HFT market‐taking activity, aggressive high‐frequency traders tend to wipe out limit orders in the direction that they trade, increasing bid‐ask spreads and resulting in higher realized volatility from the bid‐ask bounce. Figure 5.4 shows the basic mechanics of how aggressive HFT increases bid‐ask spreads. The average proportion of aggressive high‐frequency traders in stocks varies from stock to stock, but changes little over time.

Like many recurring events in the financial markets, trades initiated by aggressive HFTs leave a specific signature in the markets, according to research from AbleMarkets.com. As a result, aggressive HFT activity can be measured and recorded in even perfectly anonymous markets, and the impact of the activity can be readily and objectively examined. Table 5.1 shows the daily average aggressive HFT participation in selected financial instruments on August 31, 2015.

While the mechanics shown in Figure 5.4 may follow all market‐taking orders, two key issues pertaining to aggressive HFT behavior may particularly exacerbate available liquidity. Aggressive high‐frequency traders tend to execute bursts of market orders at once, potentially deeply affecting the liquidity on one side of the limit order book. This situation may occur at any time. Another liquidity‐draining tactic occurs when aggressive HFTs act in response to major market announcements. At those times, aggressive HFTs use their fast infrastructure to reach the markets just ahead of competing institutional traders, substantially worsening execution for the latter. Avoiding trading in the same direction as the bursts of aggressive HFT for about 20 minutes allows liquidity to replenish itself. A wait‐and‐hold strategy following bursts of aggressive HFT activity can significantly help performance of execution traders.

Following aggressive HFT can also significantly improve the performance of portfolio managers. Several studies confirm the aggressive HFT impact on market volatility. Some find that aggressive HFTs are more active during the periods of high market volatility, potentially causing said volatility. AbleMarkets.com estimates that stocks with higher aggressive HFT display consistently higher volatility.

Volatility is risk. With higher volatility come higher returns, as the holders of riskier assets demand compensation for the risk in their portfolio. Risk generated by aggressive HFTs is also linked with returns, remarkably consistent across various financial instruments and asset classes. Specifically, financial instruments with higher aggressive HFT participation have higher risk and higher returns. In addition, the risk generated by aggressive HFT falls into the “instrument‐specific” or “idiosyncratic” category, and not only is it uncorrelated with that of other financial instruments, but it can also be diversified in a portfolio. Thus, if one prefers to increase returns while increasing diversifiable idiosyncratic risks, one should pick financial instruments with higher aggressive HFT participation. If, on the other hand, one is more concerned about minimizing risk, investing into instruments with lower aggressive HFT participation allows the portfolio manager to increase the Sharpe ratio of the portfolio. With reduction of aggressive HFT across financial instruments, volatility declines faster than do returns, allowing for a Sharpe ratio decrease.

The insights gleaned from understanding aggressive HFT activity in various financial instruments do not last just microseconds or milliseconds. HFT is sticky across time, as well‐performing algos do not get turned off suddenly, but may be phased out incrementally as their performance begins to wane. Due to the HFT stickiness, the volatility driven by the aggressive HFT participation detected today can persist for hours, weeks, months, and even years. In fact, even portfolio managers that choose to reallocate their investments only once a year would do much better if they included aggressive HFT participation as one of the factors in their allocation decision framework.

Incorporating aggressive HFT in a short‐ or long‐term portfolio management framework can be as easy as multiplying or dividing existing portfolio weights by a simple factor: (1 + AHFT), where AHFT is the average aggressive HFT participation in a given financial instrument over the past day, week, month, quarter, or year. The horizon of aggressive HFT averages should match your projected holding period. For example, managers reallocating their portfolios once a month would use the last month's average of aggressive HFT activity in tweaking their portfolios. Those with quarterly or annual investment horizons would incorporate the average aggressive HFT activity for the preceding quarter or year, respectively. For portfolio managers seeking extra returns at the expense of portfolio risk, the existing portfolio weights should be multiplied by (1 + AHFT) to increase allocations to stocks with proportionally more active AHFT. For portfolio managers seeking a higher Sharpe ratio, the same weights should be divided by (1 + AHFT) to scale down investments with higher AHFT exposure.

Most interesting to other traders and investors, however, is the fact that this type of insight has the power to move the markets not in microseconds, but in minutes and, sometimes, for as long as half‐hour spans. As a result, observing and following aggressive HFT strategies can lead to highly profitable results for short‐term investors and execution traders alike.

Using the insights from aggressive HFT to understand the half hour before a news event is a particularly novel finding. The following is an example. On Friday, October 2, 2015, traders worldwide watched for the news release for nonfarm payrolls by the US government at 8:30 AM. The news was worse than expected and the markets plunged over 1 percent nearly instantaneously. What role did aggressive HFTs play in the collapse of the market in response to the news announcement? Were aggressive high‐frequency traders to blame for the market's response?

According to Bloomberg, the consensus forecast for the month‐to‐month nonfarm payrolls figures was an increase of 203,000 (see Table 5.2), while the actual figures clocked in much below at 142,000, prompting Bloomberg to report the following:

Forget about an October rate hike and maybe forget about a December one too. The September employment report came in weaker than expected on all scores with nonfarm payroll at 142,000, well under the low estimate for 180,000. To seal the matter, downward revisions to the two prior months total 59,000. Average hourly earnings also came in below the low end estimate, at an unchanged reading and a year‐on‐year rate of 2.2 percent which is also unchanged. And the labor market is shrinking! The labor participation fell 2 tenths to a nearly 40 year low of 62.4 percent.

The disappointing numbers sent the US equities markets straight down. For example, the S&P 500 ETF (NYSE:SPY) dropped nearly instantaneously from $193.50 to $191.00 in a matter of seconds. However, the impact was rather temporary—as the news was absorbed throughout the day, the market improved and SPY rose to $195.30 by the end of trading on Friday, as Figure 5.5 shows.

According to AbleMarkets research, prior to the news announcement, the pattern of aggressive HFT activity oscillated from selling to buying and back to selling, as shown in Figure 5.6 for the S&P 500 ETF (NYSE:SPY). As Figure 5.6 shows, aggressive HFT sellers outnumbered aggressive HFT buyers from 8:04 to 8:14, and again from 8:24 to 8:34, although to a much smaller extent the second time around. The heightened aggressive HFT seller activity from 8:04 to 8:14 in SPY cannot rule out an insider trading activity on the then‐embargoed soon‐to‐be‐released jobs data; however, further analysis is needed to ascertain or dispute such activity.

Aside from equities, it is noteworthy to mention the euro foreign exchange rate following the jobs announcement on October 2, 2015. While most currencies and commodities had aggressive HFT buyers and sellers at comparable levels until 9:00 AM on October 2, 2015, the EUR/USD exchange rate had a distinct spike in aggressive HFT buyers from 8:31 to 8:41 AM (41 percent for aggressive HFT buyers by volume among all buy trades in EUR/USD vs. 11 percent for aggressive HFT sellers). This sudden spike in the HFT buying activity illustrates that, in response to announcements, aggressive HFTs participate in short‐term arbitrage.

Throughout the rest of the day, aggressive HFTs helped prop the market back up. Figure 5.7 shows average participation of aggressive HFTs for all the market and marketable buy orders and sell orders by volume across the Dow Jones Industrial stocks. As Figure 5.7 shows, from 9:30 AM to 4:00 PM ET, aggressive HFT buyers exceeded aggressive HFT sellers on average in every one of the stocks shown. Throughout the day, however, aggressive HFT participation varied, as shown in Figure 5.8 for a case of American Express stock.

Are aggressive HFTs to blame for the sharp sell‐off following the news? The data indicate that most likely not, since aggressive HFT activity was quite balanced after the news announcement period, with the proportion of aggressive HFT sellers exceeding that of aggressive HFT buyers by a mere 3.5 percent (38.5 percent and 35 percent, respectively). However, aggressive HFT trading on fully embargoed information at the top of the hour prior to the news announcement when the aggressive HFT sellers exceeded aggressive HFT buyers by 19 percent (34 percent versus 15 percent) cannot be completely ruled out.

AGGRESSIVE HIGH‐FREQUENCY TRADING IN EQUITIES

Aggressive high‐frequency trading (HFT) is a classification of electronic trading strategies that rely on ultra‐fast infrastructure and market orders to take advantage of news, predictive analytics, or short‐lived information asymmetries. Unlike passive HFTs that tend to provide market‐making services, aggressive HFT models attempt to reach the markets prior to others to capitalize on short‐term market inefficiencies. AbleMarkets.com Aggressive HFT Index tracks aggressive HFT activity in real time and has developed statistical insights into aggressive HFT behavior, some of which are summarized here.

Among the S&P 500 stocks, for instance, aggressive HFTs are more prevalent in equities with

• Higher prices
• Lower dividend yield
• Higher volatility, measured as a standard deviation of daily returns

On average, a $100 difference in stock prices attracted 3 percent more aggressive HFTs by volume. Not surprisingly, Google (NASDAQ:GOOGL) is the stock with the highest average participation of aggressive HFTs registering close to 38 percent average daily aggressive HFT participation in 2014. With a stock price well over $500 per share at the time this chapter was written, Google is one of the most expensive issues in the S&P 500. An explanation for the phenomenon may lie in the value of fixed costs, such as a bid‐ask spread, of trading using market orders relative to prices of stocks: for high‐priced stocks, the spread and other fixed costs account for a smaller percentage of the price, allowing the traders to keep a larger share of the gains.

Stocks with lower dividend yields also attract aggressive HFTs. On average, a 1 percent decrease in dividends accounts for a 1.1 percent increase in aggressive HFT participation and explains 8.5 percent of variation in aggressive HFT participation among the S&P 500 stocks. Companies paying high dividends tend to be mature businesses and may detract aggressive HFTs seeking a volatile environment. This suggests that firms may be able to manage aggressive HFT participation in their stocks by adjusting their dividend policy, in conjunction with other factors.

Stocks with higher volatility also have a higher proportion of aggressive HFTs. A 1 percent increase in volatility measured as an annualized standard deviation of daily returns based on closing prices translates into a 0.23 percent increase in aggressive HFT participation.

It is not immediately clear, however, whether aggressive HFTs seek out high volatility, whether aggressive HFT participation induces higher volatility in stocks, or both. However, a 1 percent increase in aggressive HFT participation translates into 0.17 percent in additional annualized volatility across all the stocks in the S&P 500 index. Overall, differences in aggressive HFT participation account for 2 percent of variation in volatility among all of the S&P 500 stocks. Aggressive HFT participation is, therefore, a highly predictive metric of volatility: in comparison, the celebrated and often‐used GARCH model only accounts for 5 percent of variation in volatility among the same group of stocks. With higher volatility comes higher average returns, so, indirectly, aggressive HFT contributes to better performance of stocks. Specifically, a 1 percent increase in aggressive HFT participation drives up average annualized returns by 0.23 percent among all of the S&P 500 stocks. Even long‐only portfolio managers may want to take aggressive HFT participation into account in order to fine‐tune the risk allocations in their portfolios.

As the latest research shows, both the participation of aggressive HFTs and the propensity of a stock to have a flash crash can be influenced by carefully chosen corporate actions. Before any preventative measures can take effect, however, investors can perform due diligence to ascertain their chosen stock's vulnerability and portfolio risk.

The aggressive HFT is here to stay, and understanding its presence is more necessary than ever before. Today, participation of aggressive HFTs can be readily included in portfolio and trading decisions, as well as risk management and, specifically, collateral valuation. As big data analytics and computer technology continue to proliferate in finance, the applications surrounding aggressive HFTs will only expand further.

In the last few years, a number of exchanges and dark pools emerged claiming that their businesses will exclude high‐frequency traders (HFTs) detrimental to institutional investors. Almost invariably, the HFTs in question happened to be the so‐called aggressive HFTs: HFTs that execute mostly using market orders and have been shown to erode liquidity, causing short‐term volatility in the process. Although the idea of excluding aggressive HFTs may be appealing to investors, the realities of modern microstructure preclude this from happening. Given the NBBO requirements, in the US equity markets all orders are routed to an exchange with the best available quotes. As such, all orders, whether HFT or not, are herded to the same spot at the same moment—precluding any one venue from shunning aggressive HFT. As a result, most of today's exchanges in the United States have a similar proportion of aggressive HFTs by volume of executed trades.

AGGRESSIVE HFT IN US TREASURIES

US regulators have recently questioned the role that high‐frequency trading plays in the bond market. The latest research shows that aggressive HFTs initiate, on average, 20 percent of trades in the US Treasuries market. Aggressive HFTs also often trade US Treasuries when no one else does. It accounted for nearly all of the trades on the post‐Thanksgiving Monday in 2014 and the post‐Memorial Day Tuesday in 2015. Third, the participation of aggressive HFTs in the US Treasury market has declined slightly in 2015 from 30 percent in much of December 2014 and January 2015 to 13 percent at the end of July 2015.

Figure 5.9 shows aggressive HFT participation as a percentage of volume traded for US Treasuries for the period of October 2014 to July 2015. While aggressive HFT in US Treasuries made up 20 percent by volume, it varied over time. There were clear periods of volatile aggressive HFT behavior (e.g., December 2014 to January 2015), where aggressive HFT ranged from 15 percent one day to 35 percent the next day, and stable aggressive HFT behavior (e.g., June to July 2015), where aggressive HFT participation changed by less than 1 percentage point from day to day.

The observed patterns of behavior can be due to several factors, such as news, microstructure issues, and the volatility in the participation of other non‐HFT traders. The lower is the number of non‐HFT in the US Treasury markets, the higher is the relative participation of aggressive HFTs, as documented in Figure 5.9.

Overall, aggressive HFTs do not appear to be a driving force of the US Treasury markets or to be significant enough to warrant a concern, at least not yet. By comparison, the participation of aggressive HFTs in equities markets routinely tops 30 percent in many S&P 500 equities.

Notwithstanding the current aggressive HFT levels, the participation needs to be proactively monitored to capture any further developments and to develop timely responses and regulations to aggressive HFT in US Treasury markets, the markets of strategic importance to the US economy.

AGGRESSIVE HFT IN COMMODITIES

In general, aggressive HFT participation in commodities has been growing but is still far behind equities. As shown in Table 5.1, on August 31, 2015, a typical trading day, average aggressive HFT participation in selected commodities ranged from 10 to 20 percent by volume, while average aggressive HFT in equities went as high as 40 percent.

Lower aggressive HFT by volume in commodities likely reflects the fact that commodities are still dominated by human traders as opposed to algorithmic systems that comprise a fair portion of the volume in equities.

Most commodities share another important property with equities as it relates to aggressive HFT: One day's average aggressive HFT participation accurately predicts the next day's volatility in most commodities. Thus, a rise in aggressive HFT tends to be followed by a hike in the next trading day's one‐month implied volatility. Similarly, a drop in aggressive HFT activity observed on a particular day is often followed by lower implied volatility on the next trading day. Figure 5.11 shows a rescaled version of Figure 5.10, where the relationship between aggressive HFT in crude oil and one‐month implied volatility on crude oil is a lot more obvious.

The dependency between aggressive HFT participation and implied volatility can be further verified mathematically. The relationship of implied volatility and aggressive HFT is highly persistent from one day to the next. For instance, the correlation of today's average aggressive HFT level with tomorrow's implied volatility is 38.21 percent, a number similar to the correlation of tomorrow's average aggressive HFT level with tomorrow's implied volatility. Furthermore, day‐to‐day changes in average aggressive HFT are even more predictive of the next day's implied volatility than day‐to‐day changes in the implied volatility itself. In other words, both the levels and the changes in aggressive HFT behavior can help predict future implied volatility in crude oil, and, more generally, in commodities.

In general, daily values for implied volatility closely follow the prior day's averages of aggressive HFT participation. Not shown in the study is an even higher dependency that exists between implied volatility and aggressive HFT in the intraday setting. Investors can harness levels of aggressive HFT data to predict future volatility in optimizing execution and pricing options.

AGGRESSIVE HFT IN FOREIGN EXCHANGE

The behavior of aggressive HFTs in foreign exchange is comparable to that in other instruments in the fixed income space, and in Treasuries in particular. In foreign exchange, the relative proportion of aggressive HFTs has been holding steady at 10 to 15 percent in most major currencies, such as Japanese yen, Swiss franc, British pound sterling, Australian dollar, and Canadian dollar, as Figure 5.12 shows.

Foreign exchange is the notorious wild west of an asset class. Unregulated and traditionally clubby, it is also the most liquid and voluminous market, accounting for some $1.5 trillion in trading volume per day. In the face of automation, even foreign exchange trading and portfolio management has seen a push to HFT. While aggressive HFT is still nascent in most currency pairs, it is growing. Figure 5.12 shows the evolution of aggressive HFT as a percentage of daily trading volume in major foreign exchange pairs. As the figure shows, proportion of aggressive HFT in currencies may vary significantly from one day to the next, potentially impacted by institutional flow, such as quarterly cross‐border rebalancing. In fact, the nature of foreign exchange participants may drive a fair bit of observed variation of aggressive HFT participation: a single large cross‐border flow resulting from a foreign acquisition, for example, may dominate the markets for just one day resulting in a lower relative proportion of aggressive HFT by volume. Just as in equities, commodities, and Treasuries, however, aggressive HFT in foreign exchange is “sticky” as the algorithms are tweaked infrequently and the capital allocated to the aggressive HFT tends to change slowly.

CONCLUSIONS

Aggressive HFT is an important market participant that most other participants want to track and have a set of rules to use to harness aggressive HFT behavior. Aggressive HFT is also becoming an established player, and its impact is becoming clearer and easier to quantify and analyze. All types of financial market professionals are affected by aggressive HFT, as it raises risk across all time frames and all electronically traded asset classes. Understanding aggressive HFT behavior helps market participants mitigate the impact of HFT.

END OF CHAPTER QUESTIONS

1. What is aggressive HFT?
2. How does aggressive HFT impact the markets?
3. What can investors do to harness their aggressive HFT exposure?
4. Are some exchanges more conducive to aggressive HFT than others?
5. What are the differences in aggressive HFT across different asset classes?