OUTRIDER

Flood cover for horticulture, paid on what was measured.

Parametric cover for horticulture, built on satellite radar. In development, starting in New Zealand.

No claim to fileSettlement is days rather than months. How fast it settles is something you design into the book.
21,000 parcelsMeasured across Cyclone Gabrielle from data that already existed, without installing anything on a single block.

Rain is not flood. A rainfall trigger records what fell over a grid cell far larger than any orchard inside it. It does not record whether that orchard went under water.

Two maps of the same 10 km area of the Heretaunga Plains during Cyclone Gabrielle. On the left every parcel inside the grid cell is shaded a single uniform tone. On the right the same parcels are shaded individually, ranging from untouched to entirely under water, with the flooding following the river and the low ground.
One 10 km grid cell on the Heretaunga Plains during Cyclone Gabrielle. A rainfall trigger assigns a single value across the whole cell. Inside it, some blocks went entirely under water while their neighbours stayed dry.

Outrider measures the water on the block itself: the inundation of each land parcel, read from satellite radar that sees through cloud and at night. The unit is the legal parcel, the same one a policy is written against. Parametric means the payout follows that measurement instead of an assessed loss. It all runs on free, openly licensed satellite and elevation data.

A sensor answers a different question. It measures one point, and has to be installed on every property it protects. The index measures the whole parcel from data that already exists, which is how the Cyclone Gabrielle validation covered roughly 21,000 land parcels without installing anything on any of them.

HOW IT SETTLES

01Measure Satellite radar records the share of the insured parcel that went under water.
02Read the schedule The payout schedule is agreed and disclosed before the season, not fitted after the event.
03Pay No adjuster, no proof-of-loss file, nothing to negotiate.

MEASURED INDEX

  1. Satellite measures the parcel
  2. Schedule is read
  3. Payment issued

LOSS-ADJUSTED COVER

  1. Grower reports the loss
  2. Insurer reviews the claim
  3. Adjuster travels to the block and assesses it
  4. Loss is quantified, and negotiated
  5. Payment issued on settlement

EVIDENCE

Validated on Cyclone Gabrielle against the Hawke's Bay Regional Council's survey-grade flood mapping, across roughly 21,000 land parcels. That mapping was produced by the regional council, independently of our radar measurement, and built from aerial photography, field inspection and ground contours. It maps extent, not depth, and does use optical satellite imagery among its inputs.

The method has also been tested on flood events outside New Zealand that played no part in building it.

Two things are fixed. The index, meaning the share of the parcel observed under water, and the schedule agreed against it before the season. The model that reads the index is deliberately replaceable: a new one takes over only if it beats the current one on a frozen benchmark without increasing either missed floods or payouts on parcels that stayed dry. The measurement gets better over time. What you are paid on does not change.

Correlations, miss rates, out-of-sample results and the failure modes are set out in the technical dossier, where they carry their dates and their qualifiers.

Every parametric product carries basis risk, the chance that the index and the actual loss disagree. What matters is whether anyone has measured it. We have, including where it fails: the radar band in use cannot reliably see standing water under a closed orchard canopy. That is a property of the wavelength, and better tuning will not fix it. It lands on exactly the crop we are aiming at.

WHERE THIS GOES

A flood takes the season. What it often takes with it is the orchard.

Cover in this market is built around a lost harvest, because that is what a loss adjuster can count. But water that sits on a perennial block, and the silt it leaves behind, does something slower and more expensive. We can see it in the satellite record: measured against matched blocks that did not flood, most of the first season's canopy loss on the worst-affected parcels is still visible three growing seasons later. Two things that is not. It is a measure of greenness rather than economics, so a replanted block greens up again while the grower waits years for it to bear, which means it understates the loss rather than overstating it. And it is not a dollar figure: the index has not been shown to predict what a flood costs.

The season is the insurable event. The asset is the actual loss. Almost nobody covers the second, and we think that is the product worth building.

Getting there is a measurement problem before it is an insurance one, and we would rather be plain about the order. Extent, meaning how much of this block went under, is where we are now. How long the water stayed, and what it left behind, is where the value is. Each rung has to be measured and validated before it can be priced, and we will publish where each one stands.

WHY IT DOESN'T ALREADY EXIST

Not for want of satellites. Public flood-mapping services are excellent and largely free, and we use them as a benchmark. But they were built to direct an emergency response, and that job rewards a different kind of caution: be certain that mapped water is really there, because crews and helicopters get sent to it. Caution of that kind is bought at the price of leaving some flooded ground off the map.

For a response team that is the right trade. For a payment it is the wrong one. An unmapped flooded block is a grower who paid a premium and gets nothing, which is the failure that rainfall-triggered products in this market have struggled with. Building the measurement for the payment, rather than borrowing one built for the response, is the whole idea.

It is also genuinely hard, which is the more honest reason the category is empty. Our own method still misses flooding under a closed canopy, and below a certain parcel size the measurement is too coarse to stand behind. Those limits are measured, published, and the thing we are working on.

HOW WE WORK

Every test is written down before it is run, including what result would count as a failure, and graded against that document afterwards. Four pre-registered attempts to improve on the current method were run and graded as no improvement, and were not adopted. A fifth was invalidated by a bug in our own evaluation harness, and is recorded as unresolved rather than disproven.

Numbers here have moved down as often as up. When one does, it gets corrected in public with the reason. For an audience that underwrites uncertainty for a living, we think that is worth more than a clean figure. It is the same discipline that has to hold when a payout is disputed.

WHO WE'D LIKE TO HEAR FROM

Growers and their advisers. Particularly anyone who was under water in 2023 and can tell us what the cover did and did not do.

Insurers and brokers who have looked at parametric cover here, whatever the conclusion. The ones who tried it and stopped have learned things we would rather not learn twice.

Anyone holding a flood loss record. This is the useful one. A dozen historical claims with a date, a location and an amount paid would tell us more about whether this index tracks real money than any amount of further satellite work. We would happily test against it and show you exactly where it fails.

THE BUSINESS

The category Parametric flood products have mostly triggered on rainfall. Rain causes flooding without describing it, and nobody is paid for rain. Measuring the water on the block is a different product, and the same method suits any temperate floodplain growing high-value crops. New Zealand is where it starts.
The structure An MGA, or managing general agent. It writes policies on another carrier's licence and capital, owns the underwriting decision and the customer, and cedes the risk to that capital.
What compounds Every policy written against a measured index also produces a labelled record: floods measured, and the payouts made against them. Only whoever holds the policies builds that record, and there is nowhere to go and buy one.

WHO THIS IS FOR

Insurers, reinsurers and capacity providers. The technical dossier covers the method, the validation record, the out-of-sample results and the failure modes. It is available on request, and a conversation is usually more useful than a document.

Investors looking at the category are welcome to the same material.

Get in touch on LinkedIn
FOUNDERLuca Harris
BASED INNew Zealand