Data & Analytics Insights

Sentiment as a risk indicator: when does it work best?

Dr Svetlana Borovkova

Head of Quant Modelling at Probability & Partners and Finance Professor at Vrije Universiteit Amsterdam

In this insight, part of a wider series in collaboration with Probability & Partners, we show how sentiment is a powerful risk indicator in various areas, including individual stocks, commodities, sectors, and global market indices, at different time scales. In this second insight we examine:

  1. How sentiment is most effective as an early warning signal of risk, on a global scale;
  2. The difference between slow-burn risks and sudden shocks, through three major examples – Covid-19, 2007-2008 financial crisis, the 2023 banking turmoil;
  3. The value of news sentiment signal in offering invaluable risk information.

Sentiment as an early warning risk indicator

The previous insight in this series explored the impact of major economic and political events on media sentiment in global stock markets. We observed that negative events trigger a faster and stronger sentiment response, making it an excellent risk indicator. This has been a consistent finding throughout our decade-long research on the role of sentiment in finance and investing.

Our studies have shown that sentiment is a powerful risk indicator in various areas, including individual stocks, commodities, sectors, and global market indices, and at different time scales. However, sentiment is most effective as an early warning signal of risk when underlying problems slowly intensify over a long period, becoming increasingly prominent and severe.

The onset of Covid-19 in early 2020 serves as a key example of sentiment indicators' effectiveness. During January and February 2020, sentiment indicators for global stock indices (including the S&P500, FTSE100, STOXX50, and STOXX200) showed a gradual but significant decline. This decrease was different in its speed and scale compared to typical downward trends. By February 2020, the decline had notably accelerated, coinciding with the increased mentions of words like "coronavirus" or "Covid" in news reports. However, for a long time, this sentiment decline did not coincide with any noticeable changes in the stock markets. This dramatically changed on February 19, when all major stock markets crashed amidst growing Covid-19 worries, and both sentiment and stock prices tumbled in sync.  

Figure 1: S&P500 closing price vs sentiment level, 2019-2020

News Sentiment indicator

The 2007-2008 financial crisis serves as another, more distant, example of sentiment indicators’ predictive power. The Sentiment Systemic Risk indicator (SenSR), an aggregated negative sentiment measure related to Systemically Important Financial Institutions (SIFIs), monitors the perceived risk associated with the global financial system using LSEG News Analytics. News sentiment indicator showed a gradual but consistent increase in perceived risk over several months in 2007. Other indicators used for this purpose, like the VIX or LIBOR-OIS spread, didn’t react until the collapse of Lehman Brothers, at which point the financial crisis was already well underway. A similar early warning increase in news sentiment was also observed before the European debt crisis of 2011.

Figure 2: SenSR (Sentiment Systemic Risk indicator) (blue) vs VIX (black), 2006-2014

These examples highlight the significant advantage of using sentiment signals from news sources. These signals, once they have been correctly processed and cleared of any noise, can be used to monitor the sentiment around major stock and commodity markets. They serve as an early warning system for risk, potentially alerting investors to withdraw from relevant stock markets or impacted sectors prior to any drop in price.

However, could there be instances where sentiment as an early warning indicator isn’t as effective? Consider a scenario where a negative event happens abruptly and without warning. In such situations, the sentiment signal might respond to the event (albeit very quickly, i.e. intraday) rather than anticipating it, unless there was some prior information leak. Therefore, the main distinction in sentiment’s early warning potential is dependent on the nature of the risk event: ‘slow burn’ versus ‘sudden death’.

News sentiment and the 2023 banking turmoil

The 2023 banking turmoil is an excellent illustration of this. Let us look at how news sentiment reacted to the failure of several major US and European banks in 2023. Figure 3 shows the financial sector sentiment (i.e., the negative of SenSR) throughout 2023, highlighting a significant decline in sentiment in March, coinciding with the failure of several US banks (March 8: Silvergate Bank, March 10: Silicon Valley Bank, March 12: Signature Bank, and March 15 and 19: looming failure and subsequent takeover of Credit Suisse by UBS). To gain a clearer understanding, let us focus on the events of March (Figure 4).

Figure 3: Financial system sentiment, 2023

Figure 4: Financial system sentiment, February-March 2023

The downturn in sentiment within the financial sector started on March 1, 2023. By March 3, sentiment had declined so dramatically and rapidly that the impending banking turmoil was undeniably evident. This served as a valuable signal for investors to decrease their holdings in banks and other financial institutions, and for global financial regulators to pay attention. Even if there were initial doubts, by March 5, the sentiment associated with SIFIs made it glaringly apparent that a serious event was unfolding. From March 8, the decline intensified, pushing sentiment into negative territory.

Interestingly, none of the impacted banks, even a significant one like Credit Suisse, were included in this list.

So, what happened between March 1, when the sentiment started to fall, and March 10, when the first bank collapsed? On March 1, Silvergate Bank disclosed in a regulatory filing that it was at risk of losing its well-capitalised bank status and faced risks to its ability to continue operating. It’s remarkable that a regulatory filing from a relatively minor bank could influence sentiment about the global financial system, indicating a higher level of overall risk – an event that might have escaped the attention of many investors. On March 8, SVB announced a sell-off of bonds, which triggered a subsequent bank run and further contributed to the decline in news sentiment.

If one were to consider solely the banks’ share price, it could appear that the crisis only spread to the global financial system on March 13, when a significant decline in all US banks’ share prices was observed. However, the sentiment of the global financial systemhad already signalled this crisis almost two weeks prior.

The decline in sentiment began to stabilise after March 18, when President Biden made a reassuring announcement and the Credit Suisse takeover by UBS was announced. Nevertheless, it took until the end of the summer for news sentiment to return to its previous normal levels.


Both the Covid-19 pandemic and the global financial crisis demonstrate the value of news sentiment signals in detecting “slow-burning” risks on a global scale, often well before they become apparent in other economic and financial indicators. The sentiment often declines weeks or even months before the stock markets react to these risks. On the other hand, abrupt and unforeseen risks can be detected in sentiment just a few days beforehand, if not at the time of the event, as demonstrated by the 2023 banking turmoil. Recognising this distinction when tracking sentiment related to financial markets is crucial. However, in both cases, sentiment offers invaluable risk information, albeit slightly later in the latter case.

Read more about

Stay updated

Subscribe to an email recap from:

Legal Disclaimer

Republication or redistribution of LSE Group content is prohibited without our prior written consent. 

The content of this publication is for informational purposes only and has no legal effect, does not form part of any contract, does not, and does not seek to constitute advice of any nature and no reliance should be placed upon statements contained herein. Whilst reasonable efforts have been taken to ensure that the contents of this publication are accurate and reliable, LSE Group does not guarantee that this document is free from errors or omissions; therefore, you may not rely upon the content of this document under any circumstances and you should seek your own independent legal, investment, tax and other advice. Neither We nor our affiliates shall be liable for any errors, inaccuracies or delays in the publication or any other content, or for any actions taken by you in reliance thereon.

Copyright © 2024 London Stock Exchange Group. All rights reserved.