… AND THE CROSS-SECTION OF EXPECTED RETURNS Flashcards
According to the study, what t-statistic threshold is suggested for a newly discovered factor to be considered significant?
The study suggests that a newly discovered factor should have a t-statistic exceeding 3.0, corresponding to a p-value of 0.27%.
Why is it argued that a t-statistic of 2.0 is no longer appropriate, even for factors derived from theory?
It is argued that a t-statistic of 2.0 is no longer appropriate because it doesn’t sufficiently account for the increased risk of false discoveries due to multiple testing.
What is the main idea behind using a statistical framework to address the issue of multiple testing?
The main idea is to set a higher threshold for significance (e.g., lower p-value or higher t-statistic) to reduce the Type I error rate, but this may increase the risk of Type II errors.
What is the out-of-sample test, and why is it considered the best method for mitigating the issue of multiple testing?
The out-of-sample test involves checking if a factor works on a sample period after its discovery. It’s considered the best method because it assesses a factor’s performance in a new dataset, reducing the risk of false discoveries due to data snooping.
How does the Type I error rate change as the number of tests increases?
As the number of tests increases, the Type I error rate, which represents the likelihood of falsely rejecting a true null hypothesis, also increases.
