De Econometrist neemt een statistische kijk op de wereld.
In the recently published book “Flash Boys”, author Michael Lewis highlights what he calls the most dangerous affliction on Wall Street: high-frequency trading, or in short: HFT. Lewis argues that the stock market is rigged, in favor of a relatively new type of investor that earns money not by “sound investment decisions” but through abusing sheer technological advantage over “the small investor”. What is worse, Lewis says, is that this money is skimmed off the profits of regular, hard working investors. All of this is done without a human lifting a finger.
The book was featured in an episode of the American television show “60 minutes” in which the author further explained his negative point of view on HFT. This was accompanied by a lot of media hype surrounding the subject. Financial regulators also have jumped the anti-HFT bandwagon, discussing regulations which range from heavy taxes to downright banning the practice of high-frequency trading. Even the FBI has involved itself, announcing a thorough investigation looking into HFT firms, just 24 hours after the episode of “60 minutes” with Lewis aired.
Not just the media and authorities are scrutinizing HFT, also major financial players have vocalized concerns about high-frequency trading. A leading example of this is a recent press release by Goldman Sachs, in which they distance themselves from the practice, which they have partaken in for many years before. According to Goldman Sachs, HFT is damaging the financial markets. Other financial institutions have called HFT “unfair” and have accused it of rigging the market.
In order to see if HFT is indeed damaging the stock market, it is important to take a look at the exact activities considered to be high-frequency trading and to also analyze the effect that these activities have had on the market.
In “Flash Boys”, the book mentioned earlier, HFT is explained as a trading strategy that relies on super-fast computers, high-speed data networks, and complex algorithms to make money in the stock market, simply by being faster than other traders. This definition is not entirely correct. HFT involves making money by capturing opportunities which exist in a small window of time using computer programs or algorithms. This does not imply being faster than other traders necessarily. In practice however, being faster than other investors does play a role as in capturing these opportunities one has to outspeed any competition trying to do the same.
HFT serves as an umbrella term for a number of different trading activities, all of which involve the high speed algorithmic trading. The most common example is market making. Market making is the act of offering bid and ask prices for a certain asset in order to facilitate the trading of a security. A market maker accepts the risk of holding a certain number of shares and is obliged to act on its quotes for bid and ask prices if an order is made. This can be done through buying or selling from its own inventory or by finding an offsetting buy or sell order. The market maker profits from the transaction through a small difference between the bid and ask price, called the spread. Market making has been a part of stock markets for a long time. HFT has allowed for market making between different exchanges and lightning fast adjustments of bid and ask prices.
Another common trading method employed by HFT firms is that of statistical arbitrage. This involves finding pricing inefficiencies between securities using mathematical modeling techniques. This is not without risk. Just like market making, statistical arbitrage has been around long before HFT. HFT took statistical arbitrage to new heights, by allowing for lightning fast adjustments to information. Even the processing of this information is done by algorithms analyzing digital media and news, data mined off the internet. An example of this can be found in the instantaneous trading reaction to a tweet containing a message about a bombing of the White House.
In “Flash Boys”, the author singles out one activity that can be associated with HFT firms, namely that of front-running. When a large bank, hedge fund, or other institutional investor decides they want to execute a trade, they place an order for a certain stock for a certain price and send this order to an exchange. Front running is when the receiver of the order, who can recognize that a large amount of this stock is about to be purchased, acts on this information by buying the stock before the order is processed and then sells again for slightly higher price to the original investor who sent out the order.
Front running is not new to HFT. As long as there have been stock markets, there have been people attempting to front run orders. The only difference is that in the past, these market makers were actual people manning the trading floor. Before HFT took over, front running was viewed as a frowned upon practice. There are no policies known that were successful at combating frontrunning in pre-algorithm times.
Although the trading strategies employed by high-frequency traders have been around as long as the stock market exists, HFT is a relatively new phenomenon that came with recent technological improvements in the financial sector. Its development has been rather fast and its advancements have changed the financial system substantially. In what way has HFT actually affected the market and when did that happen?
The speed at which information can travel has been important ever since the dawn of the financial system. A famous example of this is the usage of carrier pigeons to relay information in the 17th century, which allowed investors to arbitrage prices of securities across country borders before their competitors.
In 1983, financial data and media company Bloomberg launched its first computerized system to provide real-time market data to Wall Street firms. The sheer speed of exchanging information and making financial calculations was unrivalled and gave users a considerable competitive advantage. These sorts of advancements in technology have been shaping the landscape of the trading floor, accompanied by new discussions on regulation.
As an example of these anti-speed regulations was that electronic orders had to be processed by keyboard, to ensure no company could gain an “unfair” advantage by being able to process orders faster. Being a fast typist was never as valuable, but it was not for long until companies employed machines which were able to physically punch in all the keys needed to make a certain order, faster than any human could. This example illustrates the difficulty of trying to come up with proficient regulations in the face of technological advancement.
The U.S. Securities and Exchange Commission (SEC) authorized fully electronic exchanges in 1998. These exchanges made it possible for anyone with a computer and an internet connection to trade financial securities. This development consequently paved the way for computerized High Frequency Trading.
The first HFT firm was created in the garage of a college statistics professor, Jim Hawkes, from the College of Charleston. Hawkes was planning to turn a profit by using predictive formulas designed by his friend David Whitcomb, who taught finance at Rutgers University. Whitcomb’s formulas were turned into computer code by Steve Swanson, a 21-year-old computer nerd who was a former student of Hawkes, who spent the whole summer of 1989 in Hawkes’ garage writing the first ever HFT algorithms. This resulted in a system that could predict stock prices thirty to sixty seconds into the future and could automatically jump in and out of trades. Steve Swanson, who was a Star Trek fan, named this system BORG. The name stood for Brokered Order Routing Gateway. Ironically, in Star Trek, the Borg is an evil alien race that absorbs other alien species into its cybernetic hive mind.
BORG’s first preys were the market makers on the exchange floors who manually posted offers to buy and sell stocks with handwritten tickets. Not only could the algorithm better predict the stock prices, it was also able to execute a trade in just a couple of seconds. Whenever a stock price changed, BORG would trade on offered bid and ask prices by the market makers, before they could adjust them to the price change. Then, moments later, when the offered bid and ask prices adjusted to the new stock price, BORG would buy or sell back to them at the correct price.
The system made on average less than a penny on every trade it made, but as it was trading hundreds of millions of shares a day, it was a matter time before the profit was substantial. They formed an official firm, Automated Trading Desk or ATD in short. This accounted for the start of a new industry.
A few years later, ATD was not alone anymore. Other companies, both startups and established firms had entrenched themselves in the area of automated computerized trading. By 2005, HFT accounted for thirty five percent of all equity trade volume. In 2010, this number had increased to fifty six percent of all trade orders. The speed of trade execution started to be of more and more importance, as being slower than a competitor could make the difference between being on the winning or losing side of a transaction. Trading firms were rushing to automate. In 2007, Citigroup, an American multinational financial services corporation, bought ATD for $680 million.
By 2010, HFT accounted for sixty percent of all U.S. equity volume and seemed to flourish at a rate such that it seemed it would soon swallow the rest. This did not happen. From 2011 onwards, HFT has been on the decline. In 2009, high-frequency traders moved about 3.25 billion shares a day. In 2012, it was 1.6 billion. Not only did the trade volume decrease, HFT firms profit has also been diminished since its heyday. In 2009, the entire HFT industry made around $5 billion trading stocks. In 2013, this was around $1.3 billion before expenses. Today, HFT companies seem to disappear just as fast as they appeared. Its most notable case might be that of Goldman Sachs selling its HFT department recently. This department, which it acquired in 2000 for $6.5 billion, yielded them a mere 30 million this present day.
As competition increased, so did the need for speed. Competing to be ever faster, HFT firms have invested in increasingly fast technology. In 2010, the speed of a transaction had decreased to milliseconds, in 2012 this became nanoseconds. For reference sake, one nanosecond is equal to one billionth of a second. Investing in a server location close to the data centre where trades are processed has also been vital for any HFT firm, as this results in a faster connection. HFT firms seem to be better than ever at making fast trades, but they are making less money doing it. Speed does not seem to pay like it used to. As firms spend millions trying to shave off milliseconds of execution times, the market has sped up but the racers have stayed even.
In the process of its evolution, HFT has changed the shape of the stock market. Before the rise of HFT, when old school brokers were the ones doing the market making on the trading floor, the difference between the buy and the sell price was about twenty-five times larger than it is today. It has been proven that HFT has had a significant role in lowering the spread. They skim less of each trade than these old school market makers did.
There are two main reasons why HFT market making gives a smaller spread than old school market making. The main reason is the risk involved. In the past, large spreads were necessary to cover the risk of making a quote which required a long time to adjust when actual stock prices would shift. As a market maker is obliged to comply with an order agreeing to its quotes, the market maker needs a premium to cover the cost of offering the wrong quote when stock prices change suddenly. HFT allows for quick responses to changes in stock price and allows for real time, accurate predictions. Using this to their advantage, an HFT market maker with quotes that are updated every second even faster has greatly lowered its risk in doing so, thus enabling its quotes to have smaller spreads.
Also, HFT is more subject to competition. During the days of the floor trading market maker, there were gentleman’s agreements in place to ensure there was similar spread given by each market maker. This price fixing allowed them to be sure of a substantial margin for their own profits. They were able to maintain such closed monopoly agreements because of the small scale of most trading floors. HFT firms that do market making have to compete on a global scale for orders, making competition fierce. Also, using HFT one can make a large volume of trades in a small period of time, making a large profit from very small spreads possible.
With the rapid decline of profits made by HFT firms and the spread of securities, it seems as if speed traders have done such a good job of closing the gap price between buy and sell prices that they seem to have a hard time wringing out some profit themselves. Another reason HFT firms are making less profit nowadays can be attributed to the industry having recognized the importance of speed and adjusted accordingly.
For an HFT firm employing front running strategies, a slow, predictable and large order, being made by an institutional investor forms a prey. With the automation of the industry, the institutional investors have combated front running by both increasing their order speed and their order strategy. One way to do this is to use algorithms to break orders into small blocks to avoid detection of the order flow. This turns front running into a game of cat and mouse and has decreased the small margin of front running HFT firms even further.
Most research literature seems to agree that HFT generally decreases market volatility (the uncertainty about the size of changes). There have been a few occurrences that show that HFT brings a new kind of volatility, namely, volatility that occurs when algorithms go berserk. An example of this can be seen in the Flash Crash. On May 6th, 2010, the Dow Jones plunged a thousand points (about nine percent) in a matter of seconds, only to recover a few minutes after. This plunge was attributed to HFT algorithms clashing in an unexpected way, causing them sell off huge amounts of stocks, only to buy them back later. In this very short time period, the price of stocks like Proctor and Gamble, went from around ten dollars to a few pennies and back in matter of seconds.
Another instance of such a crash can be seen on a smaller scale. When the established HFT market maker Knights Capital managed to burn through its capital in the time frame of approximately one hour. A newly developed algorithm had gone astray and started to buy at high and sell at low prices, losing about fifteen cents per trade. Making thousands of trades every second, this algorithm managed to shred through 440 million dollars, singlehandedly bankrupting one of the most prominent HFT market makers.
When it comes to market making, the benefits for the market are obvious. The only victims seem to be the old school market makers who were not able to keep up with modern day HFT firms. This includes both the old school floor trading market makers and more modern market makers such as Goldman Sachs, who until recently tried desperately to compete with the much smaller, specialized HFT firms in global market making. The small spread benefits the small investor and institutional investor alike.
The statistical arbitrage reacting fast to information on the market has been a point of discussion ever since the first computers were used on the exchange floors. Arguments have been made that HFT is unfair because it can react to public information so much faster than ordinary investors, it is as if this information is privately available to them, by the time other investors adjust, the price has already been corrected. Whether this is fair or not depends on the definition of a fair market. As high-speed connections to exchanges are available to everyone, it can be argued that this is a mere competitive advantage. The past has taught us that there will always be someone who has the information first. One thing is certain, a quick response to events allows for a more efficient and up-to-date market.
When it comes to front running, which is clearly a more dubious practice, it is true that HFT firms who partake in this practice are costing certain investors money. This type of investor however, is not “the small investor” as proclaimed by Michael Lewis and portrayed in the media. Only large institutionalized investors seem to suffer. The main institutionalized investors that are a target of front running HFT firms are the major investment banks, the likes of Goldman Sachs, JP Morgan, Barclays, and so on. These types of institutions make large orders rather frequently and hence the small margin HFT firms skim off their orders adds up significantly over time. Other large institutional investors, like pension funds, need not fear the HFT threat. The small margin makes no significant difference on their portfolios, which consist mainly of securities held for a longer period of time.
While it is clear that front running may not be a fair strategy, it has existed since the dawn of the stock market. Before HFT, brokers taking share of the spread in market making or committed front running on received orders was considered as the price to pay for facilitating trading traffic and a smoothly operating market. This price has decreased greatly by the coming of HFT and the market now runs more smoothly than ever. The story brought by the media about “the market being unfair to the little guy”, blaming HFT seems to completely ignore this fact, besides its twisted view about who exactly is the victim of HFT.
As argued above, the bads that came with HFT are not new to the market, whilst it has brought many goods for big and small investors alike. While actions taken to limit the effect of the more dubious practices like front running should be encouraged, one should also acknowledge the fact that the problem of front running is not unique to HFT. Regulatory measures like banning or levying taxes rendering HFT ultimately useless will be greatly unbeneficial for most market participants.
Only certain big institutions will profit from a HFT ban. Some of these big firms will develop new ways to leverage information and technological advantages over other investors. This is why the motives of those railing against HFT should be questioned. Take for instance Goldman Sachs, who suddenly, after years of partaking in the endeavor themselves, state it to be evil only after it has become unprofitable for them.
Another example of these questionable motives can be found in Lewis’ manifest against HFT.
The hero of the book “Flash Boys”, discovers “the great threat to Wall Street that is HFT” and starts its own “fair” trading platform. A large portion of the book promotes the protagonist’s “solution to the rigged market”. The character is actually solely based on the founder (who goes by the same name) of IEX, a new trading platform with measures build in against HFT. To skeptics, it came as no surprise that Goldman Sachs emerged as the new exchange’s most prominent supporter. Currently, thanks to the help of Goldman’s support, the trading platform IEX seems to be on its way to financial success
Whether the current market is a fair one, depends on one’s definition. But regulators, politicians and other motivated parties preaching they will make the market more “fair” for the small investor seem to have their arrows pointed wrongly towards high frequency trading. The market has always been rigged in favor of those with the best information, but it is actually the computers and the math that have sliced down this advantage. It is to be questioned whether these anti-HFT parties actually are moving in favor of “the little guy”. While HFT is not free of evil, it seems that the main advocates of banning it are the big traders, having to adapt their execution methods in the face of new competition. For which side you should cheer, will probably depend on which group you belong to. Investors being part of “the little guys” should not be mislead by the recent anti-HFT narrative in the media. In the current market structure, the little guy is the winner.
This article was written by Lasse Vuursteen