Today’s newsletter is open to everyone. We are always looking to improve and focus our content. Please take a moment to respond to our reader survey.
- ■ Greenlight's Secondary Peril Shift
- ■ Users Demand "Credible" Models
- ■ Atlantic Calls Converge
- ■ Risk Reads
Greenlight Breaks Out New "Large Loss" Reporting
Greenlight Capital Re’s second quarter of 2025 delivered modest underwriting volatility, but with a significant shift in how the company communicates about losses.
Management introduced a formal “large loss events” category — a new disclosure tier designed to highlight significant, but non-catastrophe losses between $1 million and $5 million.
The move marks a change from prior quarters, when Greenlight lumped all losses into either attritional or catastrophe categories.
As CEO Simon Burton explained on the earnings call:
“During the quarter, we revised our internal definition of a catastrophe loss and have started tracking ‘large loss events’ separately. This will allow us to better distinguish between attritional losses, traditional cat events, and other significant losses that don’t meet our prior cat threshold.”
Under the updated framework, catastrophe losses are now defined as individual events with net losses of $5 million or more, after reinsurance recoveries. Large loss events, meanwhile, are those between $1 million and $5 million.
While wildfire losses and severe convective storms (SCS) didn’t materially affect recent results, they were a contributor to the reporting shift.
“We continue to monitor and reserve for wildfire exposure, even where it arises indirectly in casualty lines,” Burton said. “While wildfire did not materially impact results this quarter, it remains a focus area given its potential to produce large losses.”
This places Greenlight among a smaller group of insurers publicly noting secondary perils in non-property portfolios, reflecting a more forward-looking risk management stance.
The large loss event disclosure trend is still uncommon among public insurers and reinsurers. Most carriers stick to aggregate catastrophe loss reporting, sometimes broken down by peril or geography, but rarely isolate mid-sized events in earnings communications.
By separating these losses, Greenlight is effectively creating a “missing middle” category that could provide early signals on risk creep, underwriting discipline, and emerging loss trends.
In a reinsurance market where aggregate cover pricing, climate-driven secondary perils, and inflationary claims trends are shaping underwriting strategies, this type of reporting could gain traction.
Beyond the reporting change, rthe quarter was relatively benign from a catastrophe standpoint — Greenlight reported a 0% catastrophe loss ratio. Large loss events contributed 4.6 points to the loss ratio, while attritional losses came in at 55.5 points.
Large events in the quarter included a mix of aviation and industrial incidents, such as the Air India crash, an American Airlines accident, and a California power plant fire. For the year-to-date period, management pointed to California wildfires as the primary catastrophe driver
When It Comes to Models, Brokers Rule but Science Sells
Aon's latest Global Catastrophe Risk Management Survey says that 48% of insurance and reinsurance executives are licensing catastrophe models directly, meaning the majority are relying on brokers as gatekeepers in the industry modeling view of catastrophe risk.
It is particularly pronounced among U.S. regional insurers and international carriers outside the UK/EMEA markets, where 70% of Latin American respondents operate without direct model licenses, according the survey.
For model vendors, results mean that they will need t strengthen broker relationships and ensure their platforms can support broker-led analytics.
But for insurers, while broker expertise can substitute for internal capabilities, companies may miss opportunities for the deep insights and detailed risk navigation that come from direct model access and customization.
What Really Drives Model Selection
When it comes to model selection, practicality matters.
Aon says that when choosing catastrophe models, 44% of respondents rank "reasonableness of model methodology" as the top factor, with agreement with loss experience and model transparency following close behind.
License cost ranks only fifth, while technical features like custom adjustments and integration capabilities rank near the bottom.
Rather than competing primarily on price or technical sophistication, vendors need to focus on scientific credibility, transparent methodologies, and demonstrable alignment with real-world loss experience. The emphasis on reinsurer market acceptance (ranking fourth) also underscores the importance of industry-wide adoption for commercial success.
Regional Strategies Shape Market Dynamics
U.S. insurers adopt new model releases significantly faster (65% within one year versus just 5% for global and UK/EMEA insurers), driven by smaller evaluation teams, more frequent model updates, and streamlined regulatory approval processes.
Meanwhile, global and UK/EMEA insurers invest more heavily in full-time model evaluation resources (73% versus 27% overall) and are more likely to implement complex model adjustments addressing individual hazard and vulnerability components. These regional patterns suggest vendors need differentiated go-to-market strategies rather than one-size-fits-all approaches.
Climate Change as a Differentiator
With 68% of respondents considering climate change in their capital management view of risk—rising to over 90% among global and UK/EMEA insurers—climate modeling capabilities are becoming table stakes rather than nice-to-have features. However, only 27% incorporate climate considerations into pricing, suggesting significant room for growth in climate-informed modeling applications.
Ackman Pivots From Building to Buying an Insurer
The company is targeting insurance deals in the $1-3 billion range, with Howard Hughes maintaining control while potentially partnering with other Pershing Square affiliates for larger transactions.
CSU, TSR Converge on Above-Normal Hurricane Call
Two leading forecasting organizations have released updated predictions that paint a consistent picture of an expected busier-than-normal Atlantic hurricane season with heightened storm activity through fall.
Both Colorado State University (CSU) and Tropical Storm Risk (TSR) have converged on nearly identical forecasts in statements released last week. Each predicts 16 named storms, 8 hurricanes, and 3 major hurricanes for the remainder of the current season.
The 30-year average calls for 14.4 named storms and 7.2 hurricanes annually.
For U.S. coastal residents, both forecasts anticipate 4 tropical storms and 2 hurricanes making landfall before the season ends in November of this year.
Several key factors are driving expectations for continued activity:
Warm Ocean Waters: Sea surface temperatures across the tropical Atlantic and Caribbean remain well above normal, providing the fuel tropical storms need to develop and intensify. TSR notes that these waters have warmed significantly over recent weeks as trade winds have weakened.
Favorable Upper-Level Conditions: Both organizations expect upper-level wind patterns to become more conducive to storm development, with CSU highlighting the potential for reduced wind shear in the coming weeks.
ENSO Neutral Conditions: The absence of El Niño conditions removes a typical inhibiting factor for Atlantic hurricane activity, with cool-neutral conditions potentially providing a slight boost to storm formation.
Different Perspectives, Same Outcome
Despite reaching similar conclusions, the two forecasting teams show notably different confidence levels.
CSU expressed "higher-than-normal uncertainty" due to conflicting signals, particularly strong Caribbean wind shear that has limited activity through July. They're taking a more cautious approach, noting the mixed signals between favorable sea surface temperatures and unfavorable wind patterns.
TSR, meanwhile, shows "moderate confidence" and increased certainty since their July forecast. They expect the negative factors currently limiting activity to diminish as the season progresses, pointing to weakening trade winds and improving upper-level atmospheric patterns.
Looking Ahead
CSU plans to issue bi-weekly forecasts through the peak season, while TSR considers this their final seasonal update. Both will continue monitoring atmospheric and oceanic conditions that could shift the forecast in either direction.
I built statistical models for insurance clients to assess financial risk from earthquakes and other natural disasters. We worked with massive datasets. That made me good at modeling, and that skill has translate directly into financial modeling. I can model the hell out of anything.
Patreon’s new CFO Carlos Cabrera on risk, payments and AI, CFO Dive
On Facebook, more than 6,000 people joined a private group, ODF Wildfire Risk Map Support, a base of opposition. ODF stands for the Oregon Department of Forestry, the state agency overseeing the map’s creation. One member warned that state officials would snoop around their rural properties to tell owners what to do. “Guys this is a agenda 21,” said the member, referencing the conspiracy theory promoted in part by former Fox News talk show host Glenn Beck.
How the Rapid Spread of Misinformation Pushed Oregon Lawmakers to Kill the State’s Wildfire Risk Map, Pro Publica
McKenty shared how he helped to turn GEM—and develop its software engine OpenQuake— into an open-source, community-driven initiative, literally bringing together two teams that hadn’t always shared the same spaces: developers and scientists.
How to build an earthquake model: Bring scientists and software engineers together (literally), IT Brew
