Upon further investigation I realized that the formula provided for EXK was incorrect (excess of kurtosis has a "normal" value ~0). The pandas implementation returned a value ~0.01 away from the numpy method. I believe this may be a result of the large sums being stored. To further make sure it worked I created a 10,000 point normal distribution and got the right results.
Upon further investigation I realized that the formula provided for EXK was incorrect (excess of kurtosis has a "normal" value ~0). The pandas implementation returned a value ~0.01 away from the numpy method. I believe this may be a result of the large sums being stored. To further make sure it worked I created a 10,000 point normal distribution and got the right results.