High-frequency Trading: Profiting from News

high-frequency-trading-2

High-frequency trading is a hot topic these days. After author Michael Lewis, who revealed its perils in his book, Flash Boys, was interviewed by 60 Minutes, regulators began sniffing around for signs that these ultra-fast traders are rigging the market at the expense of other investors. Practitioners argue that high-frequency trading lowers costs and makes the market more efficient. The jury is still out.

It is no secret that traders and investors are always looking for an edge in the market. High-frequency trading automates trades to profit from mis-pricings in the market before they disappear. These trades anticipate the actions of slower traders and run in front of them to take positions at a profit. It’s like running ahead to buy 10,000 widgets after finding out Walmart has a big purchase order coming, then selling them to the retailer at a profit.

Lewis’ book decries this practice as rigging the market, but proponents argue that high-frequency trading benefits the market because it reduces the bid-ask spread and makes it cheaper for investors to buy and sell securities. Whatever the case, high-frequency trading cannot be ignored: It comprises more than 50% of U.S. equities trading volume, according to a March 2014 report by the Securities and Exchange Commission.

High-frequency traders do not just look at order flows and run ahead of them to gain an edge; they also try to get ahead of market-moving news as well. Using news analytics services, they deduce  content in milliseconds to anticipate market movements and trade on them. These trades are the subject of a paper titled, “Media-Driven High Frequency Trading: Evidence from News Analytics” by Wharton finance professor Donald Keim, INSEAD banking and finance professor Massimo Massa and INSEAD doctoral student Bastian von Beschwitz.

The authors found out that the use of lightning-fast news analysis tools, while legal, combined with high-frequency trading, can have a “significant” impact on stock returns and trading volume beyond the influence of the news content itself. The presence of these analytical tools actually amplifies market reaction both to the upside and downside, according to the paper.

Is this a cause for concern? Not so, the researchers note. On the contrary, “our results have normative implications relating to recent discussions about the regulation of high-speed sources of information and the effects of algorithmic trading,” the authors write. “They show that news analytics allow the market to incorporate information more quickly and be more efficient.”

What muddles the national conversation about high-frequency trading is the batch of bad news around it. There was the infamous “flash crash” in May 2010 when high-frequency trading helped cause the Dow Jones Industrial Average to plunge about 1,000 points, or 9%, only to recover minutes later. In April 2013, a mini-flash crash occurred after it was wrongly reported that there was an explosion at the White House. Then there was Thomson Reuters’ controversy over an influential consumer survey that was released to high-frequency traders two seconds earlier than to anyone else.

Investors who are more resourceful at finding and analyzing the news have always had an edge. “It’s just that it is now done at faster speeds.”

Thomson ended the practice after the New York Attorney General’s office put pressure on the company in July 2013. The case is “analogous to a stock market analyst with valuable information about the future value of a stock who selectively releases that information early to preferred clients,” Keim says. “Such unfair and selective release of information is quite different from monitoring customer order flow — as well as other publicly available data — and using your speed advantage to trade ahead of their execution.”

Their paper does not look at such unequal, and possibly unlawful, practices. Instead, the authors studied the effect of using ultra-fast analytics tools to interpret news that was released at the same time to the public. They found that when high-frequency traders use the service and act on the news, stock prices reacted faster as well, making the market more efficient.

While that means only traders who can afford to pay for these services enjoy this advantage, Keim notes that investors who are more resourceful at finding and analyzing the news have always had an edge. “It’s just that it is now done at faster speeds.”

300 Milliseconds

In the paper, the authors used data from a news analytics service called RavenPack. The company employs computer algorithms to analyze all the stories on the Dow Jones Newswire to see which companies are mentioned and whether they are facing good or bad news. The time it takes from the release of the Dow Jones story to RavenPack issuing its metrics is 300 milliseconds or 0.3 seconds — too fast for a human trader to match.

Keim and his co-authors studied data they received from RavenPack spanning an eight-year period that ended in 2012, excluding the time around the financial crisis to avoid skewing results. The authors focused on highly relevant articles conveying important company news such as acquisitions and share buybacks, which are more certain to move the market. They ignored articles with low relevance scores, such as news that a bank is advising on a transaction. In total, the paper considered nearly 322,000 news stories.

But the authors faced a dilemma: How could they isolate the effect of news analytics tools on the market from the impact of the news itself? A company reporting good news would see its stock rise irrespective of the presence of RavenPack. But would RavenPack cause shares to go up even higher than they normally would? To tackle the problem, the researchers considered the three versions of RavenPack’s software. The latest iteration was the most sophisticated and accurate; it also scours the news for such information as names of executives and ticker symbols, to more accurately “read” the content.

“… Real news information gets incorporated into prices much more quickly these days because of technological advances.”

Since the older software versions were not as advanced, some stories were mislabeled. The authors looked at the difference between market reaction to the mislabeled stories and articles that were correctly identified to isolate the impact of news analytics tools on the market. Importantly, the authors compared story categories without regard for the actual content. So news of a company merger could be correctly tagged as highly relevant, and it would be compared with a story about share buybacks that was actually highly relevant but wrongly tagged as having low relevance. Overall, they found nearly 25,000 articles that fit the bill.

The study discovered that the stock market reacted more quickly after RavenPack released its metrics on news articles to clients. It increased the speed of reaction by 1.3 percentage points in the case of absolute returns and by 0.5 percentage points in trading volume. After 10 seconds, 35.7% of the market’s total reaction had been incorporated into stock prices, compared with 28.4% before RavenPack came along.

“Our findings show that providers of media analytics, such as RavenPack, can have a significant impact on the market that is distinct and separate from the information contained in the news,” the researchers write. Moreover, the market impact, especially in the first five seconds in which only machines can accomplish a trade, was “positively and significantly greater” than if RavenPack were absent, according to the paper.

The study also looked at the effect of “false positives,” when a news story had been incorrectly labeled as highly relevant. The authors discovered that the market first reacted as if the story was highly relevant to the company. But after 30 seconds, enough time for a quick human trader to react, the stock price began to shift and was fully corrected after two minutes when people had more time to interpret the news correctly. “The market temporarily reacts to false positives, but then reverts quickly,” the paper notes.

The HFT Controversy

Keim says the paper set out to prove “how real news information gets incorporated into prices much more quickly these days because of technological advances.” Not only are media analytics companies able to use textual analysis to “read” newswire stories at blazing speeds and translate them into trading cues based on sentiment and importance, but high-frequency traders can act on this data at lightning speed too, he notes.

The paper’s focus, however, is different from the controversial practices swirling around high-frequency trading. Keim says the study looked at another type of ultra-fast trading – not the one attracting regulatory scrutiny. “The high-frequency traders in the news today typically are monitoring order flow at sub-second intervals and are buying and selling at lightning speed based on these observed order-flow patterns,” he points out. “In contrast, the traders in our paper are typically hedge funds which do indeed trade at high frequencies but not at the speed and frequency of those traders in Michael Lewis’ new book.”

“Market prices, in the presence of faster news analytics, will more accurately and efficiently incorporate news and information that affects the true fundamental value of the stock.”

Still, the paper discovered that news analytics services do inadvertently influence stock prices apart from the news content they are delivering to clients. But Keim says it is “unlikely” that such distortions can lead to “flash crashes” because they are uncorrelated and eventually cancel each other out. Indeed, the May 2010 flash crash was sparked by a large sell order of E-Mini S&P 500 futures contracts valued at $4.1 billion, according to an SEC report. Its downward drag on an already volatile day in the markets was exacerbated by high-frequency traders.

The use of news analytics also should not affect institutional and individual investors with longer-term horizons for investing, Keim notes. Instead, they would benefit from a more efficient market. “Market prices, in the presence of faster news analytics, will more accurately and efficiently incorporate news and information that affects the true fundamental value of the stock,” Keim says. As such, he does not see the need to regulate news analytics firms “so long as a particular service is provided to all potential users at the same price, thereby providing a level playing field.”

Citing Knowledge@Wharton

Close


For Personal use:

Please use the following citations to quote for personal use:

MLA

"High-frequency Trading: Profiting from News." Knowledge@Wharton. The Wharton School, University of Pennsylvania, 15 April, 2014. Web. 18 December, 2014 <http://knowledge.wharton.upenn.edu/article/high-frequency-trading-profiting-news/>

APA

High-frequency Trading: Profiting from News. Knowledge@Wharton (2014, April 15). Retrieved from http://knowledge.wharton.upenn.edu/article/high-frequency-trading-profiting-news/

Chicago

"High-frequency Trading: Profiting from News" Knowledge@Wharton, April 15, 2014,
accessed December 18, 2014. http://knowledge.wharton.upenn.edu/article/high-frequency-trading-profiting-news/


For Educational/Business use:

Please contact us for repurposing articles, podcasts, or videos using our content licensing contact form.

 

Join The Discussion

No Comments So Far