Analysing Bloomberg’s company news and Twitter sentiment data
R21–0888 BMAN73190 for 2021–2
For researchers, lecturers, professors and observers BMAN73190
Supporting materials
- Slides/materials: link to QA’d materials
- Requires a Bloomberg laptop/terminal/remote login
Prerequisites
- Students are expected to complete our online resource Getting started with searching Bloomberg Anywhere before engaging with the following material.
- We will assume you are familiar with the most basic concepts of using the Bloomberg terminal using either the physical or remote access options.
- Students are encouraged to complete our online resource Introduction to Bloomberg Excel add-in before engaging with the following material.
- We will use the Bloomberg Excel add-in. It will help if you have used this before too.
Brief context
- Selected Accounting and Finance Masters students are required to access social media information about specific companies in order to study the impact it may have on stock performance. (See The GameStop stock frenzy, explained by Vox, 2021.)
- Bloomberg provides access to recent and historic news and Twitter data related to specific companies.
- They also provide sentiment analysis on these items; in other words, a score of how positive or negative the content is according to an algorithm.
- Some of these data are available historically. Some are available in the Bloomberg terminal, others via the Bloomberg Excel add-in.
- (This work is also relevant to several PhD candidates as part of a bigger topic.)
Practicalities
- Group: 20
- Length: 45 minutes
- Room: online
- Discipline: accounting and finance
- Level: all
Learning outcomes
After engaging with this support, you will be able to:
- Identify and use key databases in your discipline
- Identify and use relevant specialist information common in your discipline
Suggested online resources
Session content
See the QA’d materials for Word document with images. (We would have to blur out many sections of many images before putting them out online.)
Outline
- Top level search, what is most positive and most negative recent news in the world or a region? Not particularly useful for research.
- Edit this to look at a list of companies eg S&P 500… learn how to create a custom list.
- Look at one company individually: GameStop. What kinds of reports can you get? (News sources, tweets).
- What historic sentiment data can we get, how do we download it in Excel?
Getting started with News Trends
Bloomberg describes the News Trends function:
“TREN <GO> is a news-ranking function that allows you to see which companies are generating the highest readership, the strongest sentiment, or the most news or social post volume over a selected timeframe, so you can spot trending securities and anticipate moves in the market. You can also see corresponding pricing data and recent news headlines and social posts for securities receiving increased media attention.”
We start on the ‘News Reader Activity’ tab. By default, this shows the largest increase of activity in news articles read about any company, worldwide, in the last 8 hours. (At the time the screenshot was taken, Twitter CEO Jack Dorsey had just announced his stepping down.)
- You can change this to look at the largest total news read activity.
- You can change this to look at a region of the world (Americas, Asia or Europe only).
- You can change this to look at a different trend period (between 1 hour and 1 month).
Question 1: Which security has the largest today news read activity in the last month in Europe?
Looking at news trends for a list of companies
You can also change this to look at a list of securities of your choice, instead of all companies in the world or a region. Click on the ‘Security List’ button then type in the name/ticker of a list in the amber field to the right. For example, the S&P 500 or the FTSE 350.
You can use the Equity Screening function to create lists of companies based on criteria such as sector, country or size.
Looking at social sentiment trends
We will move on to the ‘Social sentiment’ tab. Bloomberg describes this function:
“The Social Sentiment tab ranks companies by social sentiment over the selected trend period, so you can evaluate how stock performance may be impacted by social media buzz. Charts, real-time market data, and social media posts provide context for the ranking, which you can display based on most negative or most positive sentiment.”
By default, this shows the most negatively scored companies, based on Bloomberg’s own score. You can change this to look at the most positively scored companies, or adjust the region and trend period. You can hover your mouse over the elements on screen to get more details.
Because we are limited to the last month, this is not particularly useful for research.
Question 2: What company of the S&P 500 Index (SPX Index) scored the most positive social sentiment in the last 8 hours?
Getting deeper with news and social media about one company
We can find more information about individual companies. Throughout this example, we will be looking at GameStop Corp (U.S.). You can type ‘gamestop’ in the Bloomberg search bar, under ‘Securities’ select ‘GME US Equity’.
Finding company news and selecting sources
Use the ‘Individual Company News’ function to look at all news which has been attributed to the currently selected company. (Type ‘Company News’ in the search bar and select the ‘CN’ function.)
You can view the news headlines in the main part of the screen, with a code to the right to indicate the source then the date. When you have read an article, it changes colour to grey.
- You can click on ‘All Dates’ then ‘Custom Dates’ to change the period. Note, dates in Bloomberg are written MM/DD/YY. To select a specific date, choose a range that starts and ends on that date.
- You can order the results by clicking on the button just to the right of the dates. Choose from ‘Time Ordered’ or ‘Top Ranked’.
- You can click on ‘Sources’ then add ‘Twitter’ to the list of sources, then remove ‘Default Source’ to exclude everything else.
Click on the news headline to read that article. It will display differently depending on the source. For example, anything from Twitter (code TWT) will include the full tweet plus the handle.
Question 3: What was the top ranked tweet about GameStop on 31 January 2021?
You can download individual articles only.
Plotting historic sentiment changes with price changes for one company
Use the ‘News Activity Chart’ function (code GN) to “plot news sentiment and volume over intraday pricing action in a line chart, so you can visualize news/price correlations in real time”.
- The blue bars show count of news publications.
- The grey bars show last price.
- The green bars show count of positive sentiment news.
- The red bars show count of negative sentiment news.
You can change the date period from the default of last 30 minutes. If you click on the ‘1Y’ button, the frequency will change from minutes to days. Then you can enter a custom start and end date.
You can change the source from news to Twitter by clicking the Twitter button on the left.
Question 4: For the GameStop in first quarter 2021, what relationship do you think there is between change in price and volume of tweets, and does the sentiment make a difference?
To export the data behind the chart, right-click it, go to ‘Copy/Export Options’ then click ‘Copy Data to Clipboard’. You can then paste this data in Excel.
Downloading sentiment analysis scores in bulk
You can retrieve current and historic Twitter and news sentiment scores in bulk. That means, for many companies or many data items or many dates. In many cases, the Bloomberg Excel add-in provides access to more historic data than the Bloomberg terminal. You can use terminal functions to help build your Excel query.
Use the ‘Field Search’ function (code FLDS) to search for data fields for the currently selected company (still GameSpot). Under the ‘Search for Fields’ tab, click in the amber box and type ‘twitter’ then press Enter.
You can view all fields which include that term in their mnemonic or description, and their latest values. This includes ‘TWITTER_SENTIMENT’ and ‘TWITTER_SENTIMENT_DAILY_AVG’, for example. If you click on these, you get a longer description, for example, with Twitter Sentiment, you can see the range and the methodology.
You can try other searches, such as ‘sentiment’, to get News Sentiment. Make a note of the mnemonic of any fields you are interested in. Only some fields will give the latest value.
Using the Excel spreadsheet builder for historical data
Open a new Excel workbook. Click on the Bloomberg Excel tab.
Click on the ‘Spreadsheet Builder’ button.
Choose the ‘Historical Data Table’ option and click ‘Next’.
In ‘1 Select Securities’:
- Click in the box labelled ‘Search for one or more securities’.
- Type ‘GameStop’ and select ‘GME US Equity’.
- Click ‘Next’.
In ‘2 Select Fields’:
- In the box labelled ‘Search for one or more fields’, type ‘sentiment’.
- Select ‘Twitter Sentiment Daily Average’.
- Type ‘twitter’ and select ‘Twitter Publication Count’.
- Click ‘Next’.
In ‘3 Set Date Range’:
- Change the ‘From’ date to ‘01/01/2021’.
- Leave the ‘Today’ box ticked so we get data up to today.
- Leave the ‘Periodicity’ as ‘Daily’.
- Click ‘Next’.
In ‘4 Preview and Create’:
- You see a preview of the layout. You could choose to transpose axes or make other changes.
- Click ‘Finish’.
Wait for the data to load.
The spreadsheet builder has put some formulas in this Excel sheet which retrieve the data from Bloomberg’s servers. They will attempt to do this after every time you open this workbook. We recommend you copy and paste as values to remove the formulas, or export to CSV, otherwise you may lose the data the next time you open the workbook.
Question 5: When would you have to use Excel instead of the terminal?
Question 6: Which date in January 2021 has the most positive Twitter sentiment for GameStop?
Tips and good practice
- Use the green ‘Help’ key to get help about the current function. If you are on a physical Bloomberg terminal, this is the green F1 key on the special Bloomberg keyboard. Otherwise, this is a button near the top of the screen.
- To go back a screen, look for the < arrow near the top left of the screen or press the ‘Back/End’ key on the keyboard.
- You can use up to four Bloomberg windows or tabs at once. Each one keeps a record of the previous companies and functions you have recently used.
- When using Bloomberg via Excel, always start your query small, with a small range of dates and a small list of securities. Make sure your request is working correctly before building it up.
- There is a monthly download limit for Bloomberg via Excel; one limit per terminal and one limit for the whole University. We do not know when these limits are close, only once we have breached them. It is crucial that you do not download more data than you need, and do not repeatedly download the same data. Use the Bloomberg terminal wherever possible as this does not have a usage limit, for example, copy the data from a chart and paste it into Excel.
Suggested reading
This is useful for research into corporate disclosure, social media impact on retail investors and capital markets. Some references are as follows:
- Ammann and Schaub (2021, Management science) “Do Individual Investors Trade on Investment-related Internet Postings?”
- Barber, Huang, Odean, and Schwarz (2021, Journal of Finance) “Attention induced trading and returns: Evidence from Robinhood users.”
- Farrell, Green, Jame, and Markov (2021, Journal of Financial Economics) “The democratization of investment research and the informativeness of retail investor trading.”
- Harris, Neely, and Saxton (2021, Review of Accounting Studies) “Social media, signalling, and donations: Testing the financial returns on non-profits’ social media investment.”
- Jame, Markov, and Wolfe (2021, Accounting Review) “Can FinTech competition improve sell-side research quality? Can FinTech competition improve sell-side research quality?”
- Nekrasov, Teoh, and Wu (2021, Review of Accounting Studies) “Visuals and attention to earnings news on Twitter.”
- Rennekamp, and Witz (2021, Contemporary Accounting Research) “Linguistic formality and audience engagement: investors’ reactions to characteristics of social media disclosures.”
Internal ID: R21–0888