Probable maximum loss (PML)

Probable maximum loss is the largest loss an insurer, reinsurer or asset owner expects to face from a single severe event, at a stated probability or return period, rather than the theoretical worst case in which everything fails simultaneously. A "1-in-250-year PML" is the loss level expected to be exceeded, on average, once every 250 years.

PML is read directly off the loss distribution a catastrophe model produces by simulating thousands of possible events against a portfolio's exposure data, picking the loss value at the chosen return period threshold on that curve. This differs from the maximum foreseeable loss, an older, more conservative engineering concept that assumes protective systems like sprinklers or levees fail, and it differs from expected loss, the probability-weighted average across all events, which PML deliberately is not: PML describes one severe point on the distribution, not its center, unlike a portfolio value at risk figure computed the same way in finance.

PML sets reinsurance purchasing levels, regulatory capital requirements, and lending decisions on catastrophe-exposed property, since it answers "how much could we lose in a bad year" rather than "how much do we lose on average." The main pitfall is treating the chosen threshold, 1-in-100 versus 1-in-250, as interchangeable: a higher return-period PML is always larger, and comparing PML figures quoted at different thresholds without checking which was used produces a misleading sense of relative risk.