From forecasting the likelihood of a hurricane to working out thousands of random outcomes that could knock their life funds off course, insurers use sophisticated models to help them price their policies.
But floods caused by heavy rains, which are in essence random, are prompting them to turn to unusual means to work out when a dangerous and expensive downpour might strike.
Scientists at Durham University, working in collaboration with insurance broker Willis, are trying to use historical patterns of rainfall, combined with knowledge of the conditions that lead to extreme rain, to more accurately forecast when a very heavy or prolonged downpour might strike.
This would help insurers charge a price for cover that more accurately reflects the risk of paying out flood claims.
The efforts underline the increasing attention insurers, and companies that help the industry by modelling potential losses, are paying to flooding caused by heavy rain.
This played a significant part in the £3bn of claims from the deluges that hit parts of the UK last year.
Stuart Lane, professor and executive director of the Institute of Hazard and Risk Research at Durham University, said good data on rainfall levels date back only 30-40 years, when the 1960s saw a big jump in rainfall recording, usually from a series of buckets positioned on hillsides.
But to work out how likely heavy downpours are to occur, data going back much longer are needed, which means tapping more unusual sources.
“We have got to make knowledge that is normally left as inadmissible more admissible,” said Mr Lane.
Records of rainfall, for example kept by Durham, do date back to the 18th century, but Mr Lane also uses other sources, including a chronology of flood and drought events compiled by the British Hydrological Society.
Mr Lane, whose own home was flooded by the rains in September, said information can be gleaned from sources such as parishes marking the walls of churches to record how high flood waters rose.
Recorded incidents include 19th-century flooding at Pickering in north Yorkshire, when water reached almost to the top of the railway platform.
Other information includes journals kept by gentleman scientists. Even today, information collected by amateurs with weather stations in their back gardens can be valuable.
Analysing well-known flood events also plays a part. In 1947, flooding to parts of the UK meant 2m people in east London were without water for two weeks. But most emphasis, said Mr Lane, was on the damage to farmland, given the “Dig for Victory!” phase of society, which encouraged people to grow food.