Goals

• For Group Project 2, you will conduct a textual analysis and incorporate the sentiment data into the analyst report you drafted in Project 1. The ultimate goal is to determine whether non-financial. non-numeric data helps or deteriorates financial analysis when it comes to forecasting stock investment decisions.

Required

• Collect at least 4 news articles that cover earnings announcement of a company of your interest every year.
  • For example, for Amazon, you are to collect at least 40 new articles covering earnings announcement for years 2011-2020 (4 articles per year * 10 years).
  • 4 news articles per year may cover annual earnings announcement event, or they may cover different earning announcements such as for quarter 1, quarter 2, quarter 3 (quarterly earnings announcement), and annual earnings announcement.
  • The larger the dataset, the better your analysis will be.

• After collecting news articles, write Python codes that conduct textual/sentiment analysis.
  • You may refer to my lecture on textual analysis.
  • You may use (1) traditional method or (2) sentiment analysis package.

• Then, run textual/sentiment analysis and obtain sentiment scores for all news articles you collected.

• In your excel file you work on for Project 1, add one additional column for the scores you obtained from the textual/sentiment analysis.
  • Note that our financial items are based on firm-year observation. In other words, you only have one observation for one company per year.
  • For example, Amazon has one observation of total assets for year 2010 and one observation of total assets for year 2011, etc.
  • As you have at least 4 sentiment scores every year, you may calculate the average and add that to the column.

• Then, you conduct another regression analysis to predict/forecast the stock price in the future.
  • This regression analysis incorporates not only the financial items but also market sentiment revolving the company of your interest.
  • It is possible that your sentiment scores are found to be "insignificant." In that case, disregard the insignificance and use the estimated coefficient to forecast the stock price.

• Create a new word document and include the following:
  • In part 1, copy and paste one sample news article you used in the analysis
  • In part 2, copy and paste python codes you used to conduct sentiment analysis
  • In part 3, copy and paste all of the financial items data including the sentiment scores you used for the analysis
  • In part 4, copy and paste all of the regression analysis table outputs and show the steps you took to forecast stock price.
  • In part 5, discuss how sentiment scores changed your previous target stock price and whether you agree or disagree with the updated target price. Also, discuss whether the use of non-numeric, market sentiment analysis helps or deteriorates previous stock price forecasts. Discuss the advantages and disadvantages of sentiment analysis in the context of stock price forecasting.