The first thing Mangrove Property Insurance sold was not a policy. It was a theory: that Florida's homeowners-insurance market, famous for hurricanes, litigation and disappearing carriers, had changed enough to deserve fresh capital. Stephen Weinstein was not arriving from a pitch competition. The Harvard-trained lawyer had spent nearly two decades as a senior executive at global reinsurer RenaissanceRe and had watched Florida risk since shortly after Hurricane Andrew. In early 2025, he put his own money, time and reputation behind a new admitted carrier.
The bet was conditional. Weinstein said Florida's legal and insurance-loss structure had long encouraged expensive behavior. Legislative and regulatory reforms changed his mind. Without them, he told an industry interviewer, the timing would still be wrong. With them - and with a growing state full of homes that must be insured by someone - Mangrove saw an opening.
That detail matters because Mangrove is sometimes filed mentally under “insurtech,” a word that can make everything sound like an app wearing a blazer. Its actual proposition is more physical. Select the houses. Price the wind. Work through independent agents. Buy enough protection to keep paying claims after a bad season. The software is useful; the balance sheet is the product.
A familiar policy with an unusual amount of scaffolding
For consumers, Mangrove's shelf is concise. Its HO3 homeowners policy covers owner-occupied single-family homes and duplexes, with packages that can add replacement cost for personal property, water backup, screened-enclosure protection and other endorsements. Its DP3 dwelling-fire policy is for homes where the owner does not live: rentals, seasonal places and secondary residences. Claims can be reported around the clock, and the company promotes a managed repair network called Grove Repair Solution.
Those products are not inventions. That is almost the point. Mangrove is competing on which risks it accepts, how it serves agents and what stands behind the contract. It operates through licensed independent agents across Florida's 67 counties. In a category where a homeowner often needs help interpreting deductibles, exclusions and replacement-cost language, the agent is not treated as obsolete middleware. The agent is distribution, advice and local memory.
Mangrove's fast start came through Citizens Property Insurance Corporation, Florida's state-backed insurer. Private carriers can offer to assume eligible Citizens policies in a process called depopulation. In April 2025, Mangrove took roughly 30,000 policies, with an 86.4 percent acceptance rate among those selected. In June it assumed approximately 20,000 more, at an 84 percent acceptance rate. By 2026, an industry interview described roughly 65,000 policies in force, mostly from Citizens.
The clever part was the brake pedal
A takeout program can look like instant scale: regulator opens gate, carrier acquires customers, press release follows. Mangrove's telling choice was to talk publicly about restraint. It had authorization to consider as many as 81,040 Citizens policies in 2025, but said it would not exceed that aggregate. Weinstein described focusing on homes that met the company's selection criteria and could be underwritten on an actuarially sound basis within the program's pricing rules.
In other words, the machine was capable of accepting more than management wanted. This is the opposite of the usual startup tension, where distribution is scarce and every new user is celebrated. For an insurer, badly selected growth can be negative inventory: each policy may carry a future claim larger than all the premium it contributes. The first thing to fail is often not demand. It is the relationship among rate, exposure and the cost of protection.
“We'll try to not bite off more than we can chew but we also want to grow.”Stephen Weinstein, founder and CEO
Mangrove's target is the broad middle of Florida housing, not only mansions and not only pristine new construction. That creates a useful tension. The middle market is large enough to build a franchise, but heterogeneous enough to punish loose underwriting. Roof age, construction, location, occupancy and storm exposure are not decorative fields on a form. They decide whether the portfolio behaves like the model.
The $111 million that is not a funding round
The company's most eye-catching number arrived in May 2026. Buttonwood Re Ltd., a Bermuda special-purpose insurer, issued Mangrove's first catastrophe bond. The deal started with a $100 million target and closed at $111 million after demand allowed the aggregate layer to grow. This was not venture funding or money available for office chairs. Investors supplied collateral for reinsurance; in return, they earn a spread while accepting the risk that qualifying Florida named-storm losses could consume principal.
The structure provides $75 million of per-occurrence protection across three tranches and $36 million of annual aggregate protection in a fourth. The notes run for three years. Occurrence cover is built for a severe event. Aggregate cover addresses accumulation - including the third and fourth qualifying storms in a season. The distinction is Florida in miniature: one hurricane is a disaster; several can become a solvency problem.
Conceptual, not to scale. Actual recovery depends on contract terms, attachments, retentions and qualifying losses.
Mangrove says its broader 2026 program included 37 AM Best-rated or fully collateralized participants, nearly 40 markets in total, aggregate protection for later events and coverage extending to a modeled 1-in-175-year storm. Names on its public panel include Swiss Re, Hannover Re, RenaissanceRe, Everest, PartnerRe, QBE and Hiscox. A first-time sponsor getting all four cat-bond tranches priced at the low end of guidance or below was a vote of confidence from specialist investors - not a promise that storms will cooperate.
In August 2026, Mangrove added Grove Re, a wholly owned Bermuda Class 3A reinsurer. The vehicle gives the group another place to align capital with its underwriting, improve risk-transfer efficiency and potentially work with third-party capital later. This is niche machinery, but it points to the company's ambition: not merely to sell policies during a favorable year, but to remain a Florida carrier across market cycles.
What another founder can copy
Nobody reading this should copy the literal risk without an actuary, a regulator and several adults from the reinsurance market. The transferable ideas are simpler.
- Wait for the mechanism, not the mood. Mangrove entered after reforms changed the economics Weinstein cared about.
- Borrow distribution before building an audience. Citizens takeouts and incumbent independent agents created reach without a national ad campaign.
- Make selection the growth engine. The company chose policies against stated criteria instead of treating every available account as revenue.
- Finance the failure mode first. Reinsurance, aggregate cover and a cat bond address the event that can break the business.
- Explain the invisible product. Publishing the reinsurance panel turns back-office capital into something agents can discuss with customers.
There is also a service lesson. Mangrove launched a redesigned, mobile-friendly website before the 2026 hurricane season, with faster paths to claims, storm information and agent resources. That sounds mundane until power is out, rain is inside and thousands of people are trying to file at once. Customer experience in catastrophe insurance is not confetti at checkout. It is whether a policyholder can reach the carrier when the neighborhood looks unfamiliar.
When the Mangrove model stops working
- If rates cannot support the modeled loss and reinsurance cost.
- If legal reforms reverse and claim severity or litigation expenses climb again.
- If rapid growth outruns claims staffing, repair capacity or agent service.
- If selection criteria miss correlated exposure across the peninsula.
- If multiple catastrophes exceed the protection actually purchased.
Where Mangrove fits
Mangrove sits between Citizens, the necessary state-backed safety valve, and a field of established and recently formed private carriers. Its alternatives include Slide, American Integrity, Florida Peninsula, Southern Oak, Universal and other companies approved to remove policies from Citizens. National insurers participate selectively. Reciprocal exchanges offer another structure. Homeowners do not experience that market as a neat grid; they experience quotes, inspection requirements, deductibles and a letter saying their policy may move.
The differentiation, then, is not “insurance but online.” It is a concentrated operating thesis. Florida only. Independent agents. Middle-market homes. A portfolio assembled with an underwriter's veto. Risk capital from traditional reinsurers and insurance-linked-securities investors. Claims infrastructure intended to dispatch repair help rather than rely entirely on the open market after a storm.
The cost is substantial even when the company does not disclose every dollar. Weinstein said he risked a meaningful amount of personal capital. Reinsurance absorbs a large share of premium; Demotech places Mangrove's utilization measure in the 80-to-90-percent range. Cat-bond investors must be paid spreads. Teams, models, reserves and regulation all arrive before the weather. This is not a cheap way to discover whether the thesis works.
Nor is the first year a verdict. Mangrove has passed several early tests: authorization, a financial-stability rating, tens of thousands of policy transfers, a voluntary product launch, a broad reinsurance placement and a debut cat bond received well by investors. The consequential exam is the one no carrier schedules - a difficult storm season followed by fair, fast claims and enough capital left to renew the book.
For now, Mangrove's most credible claim is modest and useful. It did not find a way around Florida risk. It built a company that tries to see the risk clearly, say no when necessary and pay other sophisticated balance sheets to stand beside it. On this peninsula, that is not a slogan. It is the work.