AmCoastal's $25.5M Cat Bond Buys First-Event Retention Down to $23.5M

A retention buy-down adds nothing to the top of a reinsurance tower. It removes loss from the bottom of the cedant's own balance sheet, and American Coastal Insurance bought $25.5 million of that bottom on August 1, 2026, through a single tranche of discounted Armor Re II Ltd. Series 2026-2 notes attaching at $23.5 million of Florida named storm losses (Artemis Deal Directory, August 2026). The price was roughly $8.4 million, a rate on line approaching 33%.

Key Takeaways

• $25.5 million of limit sits directly on the retention, attaching at $23.5 million and exhausting at $49 million on an indemnity, per-occurrence Florida named storm basis, for a term running August 1, 2026 to the end of May 2027.
• A 32.9% rate on line is 2.2 times the 14.75% spread American Coastal paid on its own Armor Re II 2026-1 Class B notes in April, and 6.6 times the 5.00% spread on the Class A tranche of the same deal.
• $0.54 per share of first-event loss removed for roughly $0.18 per share of premium, against $7.21 book value and Q2 core income of $0.33 per share. Retention falls from 14.4% of equity to 6.9%.
• Second-event retention of up to $25 million, raised from $18.5 million at the June renewal, now exceeds the bought-down first-event retention. The notes carry no reinstatement.

Decomposing the Buy-Down

Armor Re II Ltd. issued a single $25.5 million tranche of Class A notes, the eighth Armor Re catastrophe bond and the sixth with American Coastal as sole sponsor. The trigger is indemnity, the peril is Florida named storm alone, and the basis is per occurrence. The notes are structured as discounted, or zero coupon, instruments rather than the spread-paying form the sponsor used in April, so the cost is funded at issuance instead of accruing as a coupon.

Attachment sits at $23.5 million of losses and exhaustion at $49 million, a band that maps exactly onto the retention the June renewal created. That retention was not a single balance-sheet position. It comprised $26.5 million held by American Coastal and $22.5 million by its captive, against $29.75 million in total the prior year (Artemis, 2026 mid-year renewal). The securitized band is marginally wider than the captive's entire participation, so however the recovery is allocated internally, the group's net exposure to a first Florida storm falls to $23.5 million.

Chief executive B. Bradford Martz described the logic without dressing it up. "Our reinsurance strategy is to buy more protection when pricing is cost-effective and retain more risk on our balance sheet when it's not," he told analysts (American Coastal Q2 2026 earnings call, August 6, 2026). About $4 million of the $8.4 million is expensed as ceded earned premium from August through December, with the remainder falling between January and May 31.

Measure	2025/26	2026/27 at June 1	2026/27 after August 1
First-event retention	$29.75M	$49.0M	$23.5M
First-event limit	$1.33B	$1.68B	$1.68B
Second-event retention	$18.5M	up to $25M	up to $25M
Total occurrence limit	n/d	$1.918B	$1.918B
Ceded program cost	$201.85M	$179.5M	~$187.9M (computed)

Read across the bottom row, the year's real trade becomes visible. The soft June 1 Florida renewal handed American Coastal $22.35 million of savings on a bigger tower; the August buy-down hands $8.4 million of it back to remove the lowest and most frequently hit slice of net exposure. Net of both moves the company pays 6.9% less than last year for 26.3% more first-event limit, with a retention 21% below where it stood in 2025/26.

What a 33% Rate on Line Says About the Bottom of the Tower

Dividing $8.4 million of premium into $25.5 million of limit gives a 32.9% rate on line. The useful comparison is not the wider market but the same sponsor's own shelf four months earlier. In April, Armor Re II 2026-1 placed $200 million across two tranches: $100 million of Class A at a 5.00% spread on 0.44% expected loss, and $100 million of Class B at a 14.75% spread on 4.22% expected loss, attaching at $225 million (Artemis Deal Directory, April 2026). Per dollar of limit, the August band costs 2.2 times the Class B layer and 6.6 times the Class A layer.

No expected loss was published for the 2026-2 notes, but the April pricing supplies an anchor: Class B cleared at 3.50 times its modeled expected loss. Applying that multiple to a 32.9% rate on line implies an expected loss near 9.4% on the bought-down band, more than double the Class B figure and consistent with a layer a moderate Florida landfall would exhaust outright.

The inference is rough, since a four-month risk window inside a discounted note is not the same instrument as a three-year spread-paying tranche and investor multiples widen as layers move down. It is still the closest visible market price for the bottom of a Florida commercial residential tower this year.

That is the number a pricing actuary should hold against any internal cost-of-capital charge applied to retained cat risk. It is also expensive against comparators: two weeks earlier USAA priced its first Florida-only per-occurrence tranche at a 5.75% spread on 2.51% expected loss, so a layer far higher in the tower clears at under a fifth of the cost per dollar of limit in a half-year that set records at almost $18 billion of issuance across 83 transactions (Artemis, July 2026).

The seasonal forecast did not obviously argue for the purchase. On August 6, five days after inception, NOAA raised the probability of a below-normal Atlantic season to 75%, forecasting 7 to 13 named storms and 0 to 2 majors (NOAA Climate Prediction Center, August 6, 2026). A buy-down struck at nearly a third of limit into a season the agency was steering below normal is a decision about earnings variance, not the seasonal mean.

On 47.3 million shares and $7.21 book value (Reinsurance News, August 2026), the trade removes about $0.54 per share of first-event loss for roughly $0.18 per share of premium, converting about half a quarter's core earnings into certainty about the first storm.

The Second Event Is Now the Expensive One

The buy-down addresses one storm. At the June renewal American Coastal's second-event retention rose to as much as $25 million from $18.5 million, a figure assuming a 1-in-100-year event followed by a 1-in-50-year event. After August 1 the group's second-event retention exceeds its first-event retention by $1.5 million. In a two-storm season, the cheaper event is now the first one.

The structure reinforces that asymmetry. The notes provide per-occurrence cover with no reinstatement; once a qualifying Florida named storm erodes the band, it does not return for the balance of the term. Martz told analysts the company "should remain profitable this year, even with 3 full retentions." After the buy-down that sequence reads $23.5 million, then up to $25 million, then whatever cascades below the $435 million cascading limit. One number in that sequence improved in August. The one behind it got worse in June.

The cost also lands in a net book that is contracting. Ceded premiums earned ran at 49.8% of gross earned premium in the second quarter, net premiums earned fell 11.1% to $69.7 million, and gross written premium fell 5.3% to $216.3 million on a 24% net pricing decrease (American Coastal, August 5, 2026). Adding $8.4 million of ceded premium to a book shedding net revenue pushes the ceding ratio higher in a quarter where the underlying combined ratio had already deteriorated to 68.7% from 62.2%.

What the $8.4 million buys is ten months, not a new retention level. The band expires on May 31, 2027, one day before the renewal that will reprice it. American Coastal has rented a $23.5 million retention at 32.9% rate on line; it has not moved it. Sponsors using capital markets limit against retention rather than capacity face the same renewal-date cliff, and a mid-season buy-down is the version of that trade with the least time to amortize its price.