10 Questions
What metrics were used to evaluate the model?
accuracy, positive predictive value, and percent return
Define positive predictive value (PPV).
PPV is the proportion of true positives to total predicted positives, representing how many of the investments a hypothetical investor will make a profit on.
Why is PPV important in evaluating the model?
PPV represents how many of the investments a hypothetical investor will make a profit on.
What was the range of periods in which the PPV and accuracy were the highest?
one-month to 20-week range
Why were the accuracy and PPV lower in shorter time period tests?
Due to fewer data for the model to make predictions, resulting in less accurate predictions.
What is the average yearly return in the S&P 500 from 2008-2020?
9.16%
Which tests perform the best in the secondary tests?
Tests with long periods of news data sourcing and short periods of holding stocks
What period of news data and stock holding is likely to produce good returns in a variety of markets?
Around 100 days before the prediction and holding that stock for around 100 days
What type of investors will benefit from using a period of news data of around 20-25 weeks and holding that stock for a short amount of time?
Investors looking to invest frequently
What does the methodology presented in the paper and the specific tests done show?
They are viable and far surpass the average yearly return in the S&P 500 from 2008-2020.
Test your knowledge about evaluating a stock and news data analysis model using various metrics and data time periods.
Make Your Own Quizzes and Flashcards
Convert your notes into interactive study material.
Get started for free