Kurtosis is a big problem in predictive analytics (read "all of finance"), because kurtotic processes can spit out subtly different results compared to your historical measurements. If you just measure mean and standard deviation, that only captures two properties of a distribution, and you can be blind to many issues. This notebook is a primer on skewness and kurtosis, which are the next two properties generally used by statisticians when discussing distributions.
All lectures can be found here.
https://www.quantopian.com/lectures
Also check out our global series of workshops here. Previous cities include London, Paris, Boston, San Francisco, and NYC.