About halfway through the storm-induced, two-hour interruption of the France-Iraq match, I began to think about an Insurance Technology Diary entry. It occurred to me that no matter what’s expected, something unexpected can intervene, and change the game. I then thought that sometimes, as the hapless Tunisia side learned the day before, it’s possible for management to screw things up all on their own.

That morning I’d seen an advert for a software vendor’s AI product. They claimed that using their gizmo would shave more than 20 points from users’ loss ratios. I nearly spit out my coffee.

The claim is ridiculous. Unless it’s set to reject every submission received, and therefore to drive premium income to zero, no piece of technology can have a precisely measurable impact on loss ratio. Any suggestion that it can is simply horse-hockey.

A few days later came the awful earthquake in Venezuela. That was another unexpected intervention in the order of things. We haven’t had a big seismic event since the 2023 Turkey-Syria earthquakes, and nothing really expensive since Northridge back in 1994. The Venezuela tragedy brought to mind a headline I’d seen earlier in the week: Australia and New Zealand join the softening party. The article was about reinsurance renewals down under. Cat rates are off by as much as a fifth there, but it’s been only a decade since the $24 billion Canterbury earthquakes.

That reminded me of a fine little book I was given called Beat the Clock: How the Property Catastrophe Underwriting Cycle Works. Published by the old-style independent, centenarian US reinsurance broker Holborn, and edited by the inimitable insurance scribbler Garry Booth, the book explains the infamous insurance clock sketched by Paul Ingrey, founder both of F&G Re and of Arch Re. His ‘infographic’ charts the 12 stages of the pricing cycle, from “Euphoria!” to “Crunch!” and back again.

Early the next week a webinar run by Willis Re, the oldest start-up broker in the reinsurance world, mentioned the return-period for a major Atlantic hurricane loss. Their natural-catastrophe modelling geniuses explained how they are now able to model the tail-risk impact of Atlantic hurricanes using super-sophisticated tech.

Modellers obviously remain unable to determine precisely when hurricanes will occur, what their intensity will be, or whether they’ll make landfall. They certainly cannot credibly promise to slice 20 points off your loss ratio. However, enormous increases in computing power and the application of AI have enabled them to fine-tune their engines. That has moved the discipline on enormously since a chap I used to work with would run his 10,000 Monte Carlo trials overnight (until the time the office cleaner unplugged his tower for the vacuum).

However, advanced modelling technology does not yet factor in the insurance pricing cycle explained in Beat the Clock. Whenever underwriting is profitable, capital floods into the sector, forcing prices down. It doesn’t seem to matter what the tech tells you is the correct technical price for a risk. The market floor always seems to lie a very great distance below the proverbial ‘right price’.

That’s why it’s exasperating to read that rates have fallen so far for cat covers. Perhaps prices were well above correct technical levels, but not long ago reinsurers hadn’t earned their cost of capital for years. Then, despite tools to help them price their products adequately, they charged competition-driven rates. Now, the tools are even better. But it’s happening again. The cycle is relentless.

Cat modellers can get better and better at nailing down the technically correct ROL for a chunk of Florida wind business, but the competitive environment is more important, always has been, and is today. MS Re reported this week that global cat market pricing has reached the technical floor. Howden Re has warned that reinsurer returns are likely already back to their cost of capital.

As the France-Iraq game resumed, I realised that football is  the same. A great player sidelined by bad coaching decisions will never score, and even the best players cannot deliver a result every time, regardless of the conditions. Playing against Mexico in Estadio Azteca will be a lot harder than the same pairing in Toronto.

You can have the best players, prepare them better than any others, and give them the best possible kit, but that guarantees nothing. Bad management decisions can have an even more disastrous impact on performance. Just ask the Tunisian side.

In the same way, IT cannot be certain to improve the loss ratio. The environment has a big role to play. So too does the way the tools are used.

An old friend was an historian at the University of Cambridge before he left the ivory tower. He was frustrated because no one in power ever listened to what he and his colleagues had to say. Historians brief politicians all the time, he revealed, but in his own experience, the advice of history was never followed.

Similarly, risk carriers periodically forget that losses will indeed occur, even though the tools we build may remind them constantly, and tell them how to price accordingly. Sometimes that’s disheartening, but we in the tech business mustn’t give up.

Catastrophes will happen. Losses will occur. Today’s technology can provide unprecedented insights into areas ranging from loss-experience forecasting to risk accumulation, price-based portfolio optimisation, and even the likelihood of landfalling hurricanes.

They still cannot tell us the outcome on 19 July, but we mustn’t lose faith.

Guillaume Bonnissent is CEO of Quotech.