What is an attachment point in a parametric contract?

If you work in parametric structuring, you already know the term shows up in every term sheet before the ink is dry. But it's worth stating plainly, because a surprising number of disputes trace back to someone assuming everyone agreed on what it meant.
An attachment point is the threshold a peril index has to cross before the contract pays anything at all. Below it, the policyholder absorbs the loss themselves, the way a deductible works in an indemnity policy. Cross it, and the payout structure switches on.
In a wind cover, that might be sustained wind speed at a named location hitting 96 knots. In a flood or drought cover, it might be a rainfall deficit index crossing a set number of standard deviations below normal for the season. The number itself comes from the underwriting process, built off historical loss data and the buyer's risk tolerance. What matters for this post is the mechanics: attachment is the on-switch, not the payout schedule.
Attachment point vs exit point
Where people get tangled is confusing the attachment point with the exit point, and the two do very different jobs in the contract.
The attachment point is where the payout curve starts. The exit point is where it stops climbing, the level at which the contract pays its full limit regardless of how much worse the event gets. Between those two points, most parametric structures scale the payout linearly or in steps, so a storm that clears attachment by a little pays a little, and one that clears it by a lot pays more, up to the exit point where the cap takes over.
Think of it as a ramp with a floor and a ceiling. The floor is attachment: nothing happens until the index gets there. The ceiling is exit: once the index clears it, the payout is capped at the full limit and nothing further changes the number. A contract with no exit point, or one set unrealistically high, can leave a sponsor overexposed on the tail; one set too low can leave money on the table on a severe event that should have maxed out the payout.
Setting an attachment point for a peril trigger
Setting the number means weighing loss history, basis risk tolerance, premium budget, and the reliability of the data feed that will decide the call. None of those line up on their own, so sponsors and brokers go back and forth on the number more than once before signing off.
A few things that go into the decision in practice:
- Loss history for the exposure, not just the region. A basis gap between where the index station sits and where the actual asset is located will bite you here, and it's worth pricing that gap before you fix the number.
- Basis risk tolerance. Set attachment too close to the historical loss threshold and you'll trigger payouts on events that didn't hurt the insured. Set it too far and you've built a contract that doesn't pay when it should, which defeats the point of buying parametric cover in the first place.
- Premium budget. A lower attachment point means more frequent payouts, which means higher premium. Sponsors often work backward from what they can afford to spend, then see where that lands the attachment number.
- Data source reliability at that threshold. An index built on a single ground station or one official bulletin can leave the attachment decision resting on data nobody independently checked. If the station goes down or reports late during the one event that matters, the attachment call gets made on an incomplete read, and that's exactly the moment a counterparty has room to argue about whether the trigger actually fired.
That last point is where a lot of contracts quietly expose themselves. The attachment number can be perfectly set and the structuring can be sound, and the whole thing still ends up in a dispute because nobody can agree, after the fact, on whether the peril footprint crossed the line that was written into the contract. A wide-swath satellite read of the affected area, taken before and after the event, gives both sides a record that doesn't depend on one gauge or one party's own account of what happened. That's the gap Trigger Verification was built to close: an independent observation triggered at the moment a qualifying event occurs, so the attachment question gets settled on evidence rather than on whoever argues harder.
Get the attachment point right on paper, and you've done half the job. The other half is proving, independently, when the index crossed it, before the next event forces the question. Trigger Verification exists to settle exactly that read.