News-Based Trading

News-based trading is a style of trading in which positions are taken in response to new public information, such as company announcements, economic data releases, policy decisions and breaking headlines. The trader aims to act before the information is fully reflected in the price, or to profit from the way prices adjust in the minutes, hours and days afterwards. It can be carried out by a person watching a news screen or, increasingly, by algorithms that read and act on news automatically.

The approach rests on the observation that prices do not always absorb news instantly or correctly. That sets it in tension with the semi-strong form of the efficient market hypothesis, under which public information offers no trading advantage. In practice the competition is over speed for news that is simple to interpret, and over judgement for news that is not.

• Scheduled releases. Company earnings, employment and inflation figures and central bank decisions arrive at known times. Traders prepare in advance, and the reaction depends on how the outcome compares with what was expected, not on the outcome itself.
• Unscheduled events. Takeover bids, profit warnings, regulatory rulings, accidents and geopolitical events arrive without warning. Reaction speed and the ability to judge significance matter most.
• Soft information. Commentary, interviews, analyst opinions and social media posts carry less definite content and are processed mainly through sentiment analysis.

For scheduled releases, what moves the price is the surprise: the gap between the reported figure and the market's expectation. For company earnings a simple measure is:

• $\text{Actual EPS}$ is the earnings per share the company reports.
• $\text{Expected EPS}$ is the forecast, usually the average of analysts' estimates.

If analysts expected 1.00 and the company reports 1.10, the surprise is $(1.10 - 1.00) / 1.00 = 10$ percent. A company can report record profits and see its shares fall if the market had expected more. The same logic applies to economic data, where the release is compared with the consensus forecast. Prices often rise before good news that is widely anticipated and fall once it is confirmed, a pattern summarised as buy the rumour, sell the news.

At the fastest end, firms subscribe to machine-readable news feeds in which each item is tagged and scored for computers. Algorithms compare a released number with the forecast and send orders within milliseconds, so the first adjustment of prices to a clear-cut release is largely complete before a person can read the headline. This is a branch of high-frequency trading and depends on low latency connections to both the news source and the exchange. For text that is not a simple number, natural language processing classifies the story and estimates its tone.

Slower strategies trade the adjustment that follows. The best documented is the post-earnings announcement drift, the tendency of share prices to keep moving in the direction of an earnings surprise for weeks. Others trade the pattern of returns around recurring events, such as the Fed day effect around US central bank announcements. Option traders take positions on the size of the move, not its direction, since implied volatility typically rises before a scheduled announcement and collapses after it.

Trading conditions deteriorate around news. Market makers widen or withdraw quotes ahead of major releases, so the bid-ask spread grows and slippage on market orders can be severe. Prices may jump past stop levels, and the first move frequently reverses as the detail is digested. Exchanges may impose a trading halt in a share pending an announcement, leaving positions frozen.

False and misread information is a hazard specific to this style. In April 2013 a fabricated message posted from a hacked news agency account on Twitter, reporting explosions at the White House, caused US share indices to fall sharply for a few minutes before recovering. Automated systems have also reacted to old stories that resurfaced with a new date. An algorithm that acts on headlines must weigh the credibility of the source, which is difficult to do at speed.

News-based trading uses public information. Trading on material information before it is made public is insider trading and is illegal in most jurisdictions. Disclosure rules are designed to ensure that price-sensitive information reaches all investors together: in the United States, Regulation Fair Disclosure, adopted by the Securities and Exchange Commission in 2000, bars companies from giving such information selectively to analysts or investors. Spreading false news in order to move prices is a form of market manipulation. Controversy has also arisen where paying subscribers received data releases fractionally earlier than the public, a practice several publishers ended after regulatory attention.

• U.S. Securities and Exchange Commission