Kurtosis refers to measure of the shape, more specifically the ‘peak’ or ‘curve’ of any probability distribution of a real time random variable. A Higher kurtosis generally the distribution will have a sharper peak, and longer, fatter tails. Conversely, a lower kurtosis will have a relatively flatter peak, and shorter, thinner tails.
A distribution having a negative excess kurtosis is termed as platy-kurtic (Platy = broad). The shape of such a distribution is wider, flatter peak, and thin tails, since kurtosis is low. The flatness is caused due to data being less concentrated around the mean, due to larger variations in the data set. Hence, kurtosis is also a measure of the level of fluctuation of the data within a particular data set.
Platy-kurtic or lower kurtosis indicates higher variance in the data set, whereas leptokurtic and meso kurtic are lower in terms of variance in data.