18  Peaks Over Threshold

Introduces the Generalized Pareto Distribution as the practical workhorse — model only the exceedances above a threshold, rather than the whole distribution:

def exceedances(data, threshold):
    return [x - threshold for x in data if x > threshold]

This chapter is explicitly aimed at practitioners: insurance losses, latency spikes, flood levels — the standard peaks-over-threshold workflow used across those fields, built from scratch instead of imported from a risk-modeling package.