The rebel scum, breaking norms and making your data ugly. The outliers. Even normally distributed data has them.
A mesokurtic probability distribution is one that is similar to a normally distributed dataset, given the peak-height of a random variable in statistics (think: a bell curve).
There are platykurtic distributions, which are skinnier with higher peaks than the mesokurtic distribution, and leptokurtic distributions, which are a bit shorter and fatter than the mesokurtic distribution. All these terms have “kurtosis” in them, which refers to the measure of the tails...and that includes the far-out outliers. A higher kurtosis means a wider distribution, which means more outliers (the rebel scum).
For risk managers, the higher the kurtosis, the more likely the chance that an extreme event will happen. Mesokurtic distributions have middle-of-the-road tails, while leptokurtic distributions are what keep risk managers up at night with their long tails (the distributions, not the risk managers).