How does the stock market respond to news analytics?

In recent times, the landscape of institutional trading in financial markets has witnessed a growing reliance on novel information sources, particularly ‘sentiment’ indicators derived from news wire articles. These news analytics, generated by computer algorithms, swiftly assess whether an article conveys positive or negative sentiments and holds pertinent information concerning a firm’s value. Concurrently, the proliferation of computerized trading has accelerated the process of accessing such information and the pace at which it influences stock prices. Access to this low-latency ‘meta information,’ often referred to as ‘news about the news,’ bestows a competitive edge, primarily to high-frequency and algorithmic traders like hedge funds. Nevertheless, the utilization of inaccurate low-latency signals can lead to unintended consequences when algorithms automatically trigger trades based on erroneous data. For instance, in April 2013, a misleading tweet about a White House incident resulted in a brief stock market disruption in the United States. While some swiftly pointed fingers at algorithmic trading for the reaction, others attributed it mainly to human traders.1 Regardless, it appears that news-reading algorithms might be more susceptible to misinterpreting news compared to their human counterparts. Hence, comprehending the extent of price effects associated with such meta information is of paramount importance to policymakers concerned about maintaining financial market stability.
In July 2013, Eric Schneiderman, the New York State Attorney General, reprimanded Thomson Reuters for granting select high-frequency algorithmic traders access to key economic survey data two seconds ahead of other customers. Unlike the early dissemination of economic survey data, news analytics are founded on publicly accessible news and thus represent a ‘fairly earned’ advantage. However, since news analytics facilitate swifter trading based on public information, they confer a competitive advantage akin to possessing early access to private information. The key question revolves around whether rapid trading spurred by low-latency information yields market effects distinct from the core informational content of the news. In other words, does high-frequency trading driven by news analytics induce potentially distorting price effects? Regulatory intervention, if justified, hinges on the existence of such distortions.

Media, institutional traders, and stock prices

A substantial body of academic literature has scrutinized the influence of information from traditional media sources on stock prices (e.g., see Tetlock 2007, 2011, Fang and Peress 2009, Dougal et al 2011, Engelberg and Parsons 2011, and Peress 2013). Some researchers (e.g., Riordan et al 2013, Gross-Klugmann and Hautsch 2011, Sinha 2012, Zhang 2013) has delved into the market’s reaction to news analytics but hasn’t managed to isolate a distinct impact of meta information in news analytics from the underlying news upon which it is built. In a recent working paper, we have succeeded in identifying the causal effect of meta information (von Beschwitz et al 2015). We leveraged disparities in high-frequency signals derived from Dow Jones Newswire stories between older and newer versions of Ravenpack, a prominent news analytics provider. Using corrected and back-filled signals from the new RavenPack version as a proxy for the ‘true’ informational content of the news, we observed that traders responded to the original signal released in real-time. Differences in stock market responses between the ‘true’ and real-time versions of news analytics allowed us to examine the causal impact of meta information. While such differences were relatively infrequent (approximately 3% of our sample), they occurred in sufficient numbers (24,963 cases) to conduct robust tests.
RavenPack’s signals, derived from the DJ Newswire, gauge (a) the relevance of the article to the subject firm (‘relevance’) and (b) whether the article conveys positive or negative information about the firm (‘sentiment’). Based on variations in relevance scores, we established three article classifications: High-relevance articles originally Released as being High-relevance (HRH); Low-relevance articles originally Released as being High-relevance (LRH); and High-relevance articles originally Released as being Low-relevance (HRL). HRH and HRL articles contain similar informational content, but only HRH articles were initially released as relevant to the stock market, potentially impacting stock prices more significantly. This framework enabled us to ascertain the causal effect of meta information on the stock market by comparing differences in stock price responses between HRH and HRL articles. The overreaction in stock prices to an LRH article, where no effect should exist, also indicates the causal influence of meta information.

Our research reveals that market participants respond differently to Accurate and False Negative articles, yielding three distinct effects:

1. Price Effect: The stock price response within the first 5 seconds following the release of news analytics is significantly more pronounced for HRH articles than for HRL articles. This disparity in the speed of stock price reactions amounts to 1.3 percentage points or 10% relative to the mean. Furthermore, we observed that the market overreacts to LRH articles in the short term and starts to mean-revert after 30 seconds. This pattern aligns with a causal effect of RavenPack, causing algorithmic traders to trigger an initial overreaction, subsequently corrected by human traders.
2. Volume Effect: The proportion of trade volume for the stock concentrated within the first 5 seconds, compared to the two-minute interval after release, is significantly greater for HRH articles than for HRL articles. This increased trading volume is consistent with existing theories predicting that investors with a speed advantage trade assertively on signals they can exploit before other traders (e.g., Foucault et al 2013).
3. Liquidity Effect: A stock’s liquidity measured during the five seconds following an article’s release is notably lower for HRH articles than for HRL articles. This reduction in liquidity provision by market makers and other liquidity providers is attributable to the fact that only a subset of market participants has access to news analytics, intensifying information asymmetry in the market. Our findings suggest that news analytics enhance price efficiency but concurrently decrease liquidity, raising questions about whether the resultant improvement in information efficiency outweighs the accompanying liquidity reduction.

Concluding remarks

Our study explores the impact of information provided by news analytics companies on the stock market, focusing on how this alternative delivery mechanism influences liquidity and market efficiency. This inquiry holds significance since institutional investors, responsible for the majority of trading volume in the market, rely heavily on news analytics.
Our findings underscore the considerable market impact of media analytics, distinct from the information contained in the news. These results bear normative implications concerning recent regulatory deliberations regarding the high-speed dissemination of information and the ramifications of algorithmic trading. They affirm that news analytics enhance price efficiency but potentially distort liquidity and price effects, underscoring the need for further examination of the balance between improved information efficiency and reduced liquidity.

Get back to Seikum News 🤓