Coronavirus News, Markets and AI

The COVID-19 Diaries

Pankaj Sharma

• Publisher: Routledge India
• Year: 2020
• Print ISBN: 9781032046006
• eBook ISBN: 9781000327441
• Topic: Ciencia de la computación
• Subtopic: Negocios en general

Coronavirus News, Markets and AI explores the analysis of unstructured data from coronavirus-related news and the underlying sentiment during its real-time impact on the world and on global financial markets, in particular. In an age where information - both real and fake - travels in the blink of an eye and significantly alters market sentiment daily, this book is a blow by blow account of economic impact of the COVID-19 pandemic.

The volume:

• Details how AI driven machines capture, analyse and score relevant on-ground news sentiment to analyse the dynamics of market sentiment, how markets react to good or bad news across 'short term' and 'long term';
• Investigates what have been the most prevalent news sentiment during the pandemic, and its linkages to crude oil prices, high profile cases, impact of local news, and even the impact of Trump's policies;
• Discusses the impact on what people think and discuss, how the COVID-19 crisis differs from the Global Financial Crisis of 2008, the unprecedented disruptions in supply chains and our daily lives;
• Showcases how easy accessibility to big data methods, cloud computing, and computational methods and the universal applicability of these tool to any topic can help analyse extract the related news sentiment in allied fields.

Accessible, nuanced and insightful, this book will be invaluable for business professionals, bankers, media professionals, traders, investors, and investment consultants. It will also be of great interest to scholars and researchers of economics, commerce, science and technology studies, computer science, media and culture studies, public policy and digital humanities.

Part I

THE METHOD

1

HOW TO READ THIS BOOK?

A few suggestions before you begin

To get the best out of it, here are a few suggestions:

• This book is divided into four parts. Read Part I first, as it gives vital details that will help you understand what are we trying to do and how we are doing it. Parts II and III are not mutually exclusive, but Part II is better suited for people who have some basic understanding of financial markets or those who are interested in it. Similarly, Part III is more suitable for readers with a general interest. Part IV is the gist of the key themes in this book. As such, Parts II, III and IV can be read in any order.

• Read it slowly, preferably one chapter in one sitting. Try to recall how you personally would be thinking about the markets and the world on that day and how others would be thinking on that day according to your opinion – for example, would your thinking be influenced by which part of the world you were in and so on. Think about these scenarios, and think hard. That way, you will get more out of it.

• Refer to the charts and try to understand them first without referring to the accompanying text. Make your own assessment and try to infer what these pictures are conveying. If you draw your inferences and find them different in comparison to the text we have put, then it is better because there are multiple dimensions a reader needs to really think about.

• Think about the Coronavirus Sentiment and the Market Sentiment differently and understand that these are measures of two very different things. Similarly, the sentiment on keywords will be different from either Coronavirus or Market Sentiment. Why is it important? Because it matters to understand the cause and effect. That is different from correlation.

• No subjective thing can ever be perfect because of possibilities for improvement nor can anything ever be completely right or completely wrong. Think about the linkage of sentiment with markets on those lines. The Coronavirus News Sentiment or Market Sentiment will never have a perfect mathematical relationship with actual market movement, but as long as you get it mostly right, then you will be doing fine.

A sample of our machine-aided observations

1 Oil, Again (27 April 2020), Oil’s Historic Fall: Precipitated by Quickly Worsened Sentiment? (22 April 2020), Crude and Coronavirus: Oil Futures in Negative for the First Time in History and Its Implications (21 April 2020) – We analyse the extraordinary price movements in the crude oil market. The massive destruction of demand because of the coronavirus has created a deep and unprecedented crisis in the crude oil market. On Monday, 20 April, the May United States oil futures contract went into negative territory for the first time in history. This is visible on the EMAlpha Oil News Sentiment tracker as well. While the market movements like on April 20 are related to specific reasons like storage issues, the general sentiment on the demand side is extremely negative. The oil news sentiment is at the lowest levels after the impact experienced during the dip in March because of supply glut expected as a result of the price war between Russia and Saudi Arabia.

2 Markets and Coronavirus Sentiment: Battle between Optimism and Pessimism (20 April 2020) – We are struggling to figure out until when the divergence between sentiment on coronavirus-related parametres and market performance may continue. Most markets are up by at least 15 to 20%, and we are now far away from the lows of March. However, this is only half the story. The on-ground situation on the coronavirus seems to be telling something very different. On the Global Coronavirus News Sentiment and Country-By-Country Coronavirus Sentiment, things continue to deteriorate. Except for a few exceptions like the United States of America and China, most countries are struggling to contain infections, and the statistics are plummeting. Similarly, the Global Coronavirus Sentiment has deteriorated sharply this week.

3 News Sentiment on Donald Trump and Markets (17 April 2020), Trump Losing ‘Perception Battle’ on the Coronavirus, and Why It Matters? (15 April 2020) – How did sentiment from news flow on the United States President Donald Trump quickly worsen, and why may that be important for the direction of markets? The perception that Trump has handled the ‘corona crisis’ poorly has been fairly strong in the media and logically, this could hurt the markets. On the contrary, while Trump may be doing badly on news flow sentiment, the markets have been doing rather well. After the market hit the bottom on 23 March, S&P 500 has gained 25% and during this time, (a) the coronavirus crisis has worsened significantly in the United States, and (b) the common perception is that the handling of this major health and economic challenge by the United States President has not been optimal. How do we explain this, and why do we say that Trump matters for the market?

4 Timing the Virus: Market Timing Possible with Sentiment Analysis? (13 April 2020) – Is it possible to position for market volatility in a timely manner? As our focus tends to be EM, we consider two major emerging equity markets: Brazil and India. We use our sentiment-based signals, constructed using machine translation and sentiment analysis on news flow in Portuguese, Hindi and English to trade the broad equity indices in these two countries. We construct our Brazil and India Sentiment Signals every morning at 8 AM local time, and we use it to predict the day’s market movements. More specifically, if the sentiment improved from the previous day, we go long in the market. If it deteriorated from the previous day, we go short in the market. The result, in terms of the cumulative return in excess of the equity benchmark, suggests that sentiment signals for Brazil and India seem to have navigated the recent unprecedented market volatility quite well.

5 Fed Making Data on Fundamentals Irrelevant for Markets? (9 April 2020) – The coronavirus pandemic has clearly made the United States Central Bank extremely active, and it began with an announcement of an emergency interest rate cut in the first week of March. Now, interest rates have become almost zero and the Federal Reserve is also doing big-scale asset purchases. Since 2008 when Quantitative Easing (QE) proved effective in averting a huge disaster during Global Financial Crisis (GFC), QE has become a ‘tool of choice’ for the Fed. This leads to a dichotomy in this respect. How the signals from fundamentals of the economy (such as jobless claims) will be interpreted and how markets will react to what the Fed is doing will pull the markets in opposite directions.

6 Why Local News-Based Sentiment Analysis Matters? (8 April 2020) – On news flow sentiment analysis, this question is being asked often: In this day and age of global interconnection, does local news matter anymore? Our experience with news collection for different countries in different languages confirms this. English news and translated (from local language to English) news can be quite different. It might be that analysis becomes necessary to check what is the more important driver of sentiment: is it the English news sentiment or the local news sentiment?

7 Coronavirus: Darkest Before the Dawn or No Light at End of the Tunnel? (6 April 2020) – The coronavirus news sentiment has deteriorated in 11 out of the 12 countries in our main panel (the United States of America, China, Germany, Brazil, Italy, India, Mexico, Malaysia, Norway, Colombia, Australia and Poland). This does not inspire much confidence that we are anywhere close to recovery. Also, when we do sentiment analysis for some keywords such as ‘Donald Trump’, ‘Fed Rate’, ‘Nasdaq Composite’, ‘Chinese Economy’ and ‘US Economy’, we find that, after his proactive steps over the last week or so, there has been a big turnaround in news sentiment for the United States President Donald Trump. This matters for its linkage with how quickly and efficiently the biggest economy of the world will respond to the coronavirus threat. The bad news here is that things do not look that great for S&P 500 and the United States economy sentiment.

8 How Base Rate Changes Everything (1 April 2020) – Why ‘Aggregate Market Sentiment for India’ and ‘News Sentiment on Coronavirus Sentiment for Several Countries Including India’ are showing very different patterns and how to read them to take a call on the market. The ‘base rate’ can explain this. The coronavirus in itself is a negative event, but the change in news flow sentiment would be continuous – sometimes good and sometimes bad. While the sentiment in the news about the coronavirus would keep changing, the overall impact on the market is going to remain negative. Sometimes, the negative impact is less, and sometimes it is more. The Coronavirus News Sentiment Chart only talks about this sentiment, which is in a narrow range.

9 The Dichotomy of Worse Coronavirus Situations and the Better Markets (30 March 2020) – There have been some key developments and inferences that can be drawn on the Coronavirus News Sentiment and its impact on the global, as well as country-specific, themes. While there is unprecedented coverage of the market impact of the Coronavirus COVID-19, the massive volume of this coverage makes it difficult to draw any conclusions. Related market analysis has focused on specific countries or specific asset classes or government and central banks’ responses. How do we simplify the event analysis and what are the inferences we can draw?

10 Better News Sentiment on the United States Help Emerging Markets (25 March 2020) – Over the last ten days, there have been interesting takeaways from sentiment analysis for news flow around some very important and immensely popular keywords, which matter for the United States of America. First, we discuss the key positive from this analysis. We find a little bit of improvement in the overall news sentiment on words such as ‘Donald Trump’, ‘Dow Jones Index’, ‘Fed Rate’, ‘Nasdaq Composite’, ‘Chinese Economy’ and ‘US Economy’. On a cumulative basis, the news sentiment for keywords which matter for the United States of America has improved in the last few days but only by a little. Overall, the improvement in news sentiment over the United States economy is reflecting in the United States markets as well. However, what is needed now is more confidence in Trump and the Fed on their ability to pull this off.

11 Coronavirus News Sentiment and Indian Markets (24 March 2020) – If we map the Aggregate Market Sentiment for India (not just the Coronavirus News Sentiment for the country, which will be a part of overall sentiment anyway) with the market indices, there is a clear message. The market volatility is not likely to go away until the time there is some stability in the news sentiment. The smart recovery in market performance on Friday, 20 March, when the market was up about 6%, was followed by an even deeper fall on Monday, 23 March when there was a 13% decline, gels very well with the India Coronavirus News Sentiment. The worst possible news for the global markets is that United States markets continue to remain in deep turmoil. From the behaviour of markets so far, it is likely that any turnaround will not happen until news sentiment on the coronavirus starts to get better.

12 Coronavirus, News Sentiment and Investor Behaviour (23 March 2020) – From the recent coronavirus-driven reaction of global markets, it has become quite clear that sentiment is a dominant driver of investor behaviour. In an age when the speed of news dissemination is at its peak, market sentiment requires careful consideration by investors. The initial reaction of the markets to global shocks tends to be somewhat broad-based, as all markets react to the same overall shock. However, as events evolve, local news flow sentiment also starts playing a significant role. High-profile cases lead to further deterioration in sentiment and a more negative market reaction. It is possible to use a systematic approach and use quantitative measures of sentiment to make investment decisions, and this can lead to outperformance versus the broader market.

13 Coronavirus Sentiment Deteriorating Further and What We Learn (18 March 2020) – The latest country-by-country ‘Coronavirus Sentiment’ based on news flow continues to show further deterioration in many countries including the United States of America. However, things have started to look up in China. Considering this was the origin country for COVID-19, this is a good sign. India is seeing a further fall in coronavirus sentiment and that is bearish. The bigger worry is that the cumulative sentiment, or the sum of total of news flow across all countries which we track, continues to plunge and it is very likely that, unless there is an improvement here, the global markets may not stabilise. The Indian markets continue to swing wildly, and the other important takeaway is that the cumulative market sentiment is still bearish which may not help in an immediate market recovery.

14 High Profile Cases and the Impact on Coronavirus Sentiment (15 March 2020) – On a time-series for individual countries, one notices huge fluctuations in coronavirus sentiment. In some cases in which the United States is the most prominent, the news flow sentiment has deteriorated rapidly. The same is the case with Canada in which the coronavirus news flow sentiment has deteriorated sharply. However, the reasons are perhaps different than the United States’ reasons. In Canada, this could be because of a high-profile case of infection (i.e. the Prime Minister’s wife). The variance between countries on a day-to-day basis continues; for example, Japan is the country with the most ‘negative’ sentiment on the basis of news flow for 14 March, while things look much better for Korea, Chile and Brazil.

15 Spike in Volatility and Coronavirus Sentiment (10 March 2020) – The global markets see an unprecedented spike in volatility. On 9 March in the VIX Index, the widely monitored gauge of market volatility shot up to levels that have not been seen since the Great Financial Crisis almost ten years ago. While the immediate reason was the breakdown of oil production cut negotiations between Russia and Saudi Arabia, the stage was already set for such a big market movement by the recent developments in the spread of the coronavirus. We monitor daily news sentiments in a range of emerging and developed markets. As it has become clear that Coronavirus-Related News Sentiment could continue to be a driver of market sentiment, we will continue to monitor coronavirus news and sentiment.

16 Coronavirus, Human Irrationality and Daniel Kahneman (7 March 2020) – As reflected in the reaction of global markets to the coronavirus’ threat to the global economy, it is easy to see why sentiment is often much more powerful than facts alone. This is not just limited to markets, however. Events that appeal to human emotions and our primitive instincts always have a disproportionate impact than what objective assessment would warrant. This is true for terrorist threats, fears of recession and how broad-based the impact of negative events will be, in general. In many of these areas, one of the most elegant explanations and lucid examples backed by years of research and experiments with a large sample size have come from Daniel Kahneman. In his book Thinking, Fast and Slow, Nobel Laureate Kahneman has spoken at length about ‘human irrationality’ and the way the process of thinking gets influenced by rather unexpected factors.

17 Coronavirus Sentiment Watch (3 March 2020) – The markets have reacted strongly to Coronavirus-Related News Sentiment. While it remains topical, we will try to regularly post a heat map of our country-level sentiment score for the coronavirus. China, Thailand and Iran saw the most negative sentiments, while England saw the least negative sentiment. As the daily sentiment changes for a number of countries, we believe that the sentiment score-which is an input into our systematic trading model – can be valuable in risk managing or creating new signals for any international portfolio.

18 Coronavirus and Markets: Is Local Sentiment More Important? (2 March 2020) – The sharp downward market movements across the world in the previous week has removed any doubts about the kind of impact the coronavirus would have on the investor sentiment. I...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Contents
• List of Figures
• List of Tables
• Acknowledgements
• About the Book
• Introduction
• Part 1: The Method
• Part 2: The Results
• Part 3: The Samples
• Part 4: The Inferences
• Index