When we perform a linear regression, we obtain a table with the summary results, containing some hypotheses tests, and one of them is the jarque bera, to test if the residuals follow a normal distribution (null hypothesis).
In the python documentation (scipy.stats.jarque_bera), this test should have 2000 observations. which test should we perform for a smaller number of observations?
Shouldn't the central limit theorem apply and all the residuals above 50 obs. be normal?