Leptokurtosis is observed in many financial distributions.
It means a more ‘pointy-headed’ and ‘fat tailed’ observed distribution, compared with the distributions predicted by the normal and lognormal models.
Importantly there is a fatter downside tail (‘left tail’) in the observed data.
In other words the observed frequency of large negative returns (or results) is greater than predicted - for example - by the lognormal model of the distribution assumed in the Black Scholes option pricing model.
Because of leptokurtosis, Value at Risk models which use a normal frequency distribution will understate the Value at Risk.