Retro market sources have expressed surprise at Markel Catco's initial loss estimate for 2017 catastrophe events, after it said investment returns for the year could range from a 5 percent gain to a 15 percent loss.
The firm said the forecast was based on "current loss estimates" following recent US and Caribbean hurricane events.
There is currently significant uncertainty over the possible scope of claims from hurricanes Harvey, Irma and Maria, with models showing a wide range of potential outcomes.
Markel Catco did not state the level of industry losses it was assuming in its calculations, whether it was using initial PCS data that put loss tolls below where models project ultimate losses could settle, or whether it had assumed reinsurance recoveries would limit damages.
However, multiple retro market participants said they would expect the level of catastrophe activity that has occurred in 2017 to result in more significant losses to Markel Catco's portfolio than the firm's initial projection implied.
This is based on general market knowledge of the way the company's pillared retro product functions. Its flagship retro product offers buyers up to four "pillars" of cover, or up to four payouts of their limit across a broad range of perhaps seven or eight selected risks.
A hefty premium - roughly a 70 percent rate-on-line in relation to the "pillar" size - helps to capitalise the first payout under this structure, which limits Markel Catco's losses from any one major event.
Additionally, the manager has previously said that hedges in place on its portfolio would limit the investment result from a single worst-case event to a 10 percent annual loss.
But while a major single loss is highly manageable under the pillared retro structure, a string of significant losses is more challenging as claims begin to erode capital once premium has been used to help meet a first-event payout.
The typical triggers on each risk pillar are set at relatively low-attaching levels in order to provide cover against "earnings events", and almost all buyers would have chosen to cover their US hurricane portfolios, sources said.
They estimated that most buyers would expect to claim two to four of their "pillars" in a year such as 2017 - with Florida hurricane from Irma, Gulf windstorm from Harvey or Caribbean windstorm from Maria; and, for some buyers, Mexican earthquake or other global losses all expected to trigger payments for clients with these pillars of cover.
Taking the upper end of the manager's initial loss range, a 15 percent loss to net asset value, would suggest a gross loss of around 35-37 percent, one source explained, after adding back the premium figures that would have softened the blow to income.
But if this proved to be the case, the source said this would mean Markel Catco was paying out less than two of its pillars on average to clients, which is at the low end of the scale of payouts that the market would expect.
Sources suggested that if its losses were of the scale suggested by the manager it would leave buyers questioning the value of the product, given the extent of 2017 losses.
Others expressed their scepticism in different terms, by looking at the structure in simple mathematical terms, starting from the premise of dividing the manager's roughly $4bn asset base into four $1bn pillars, comprising $700mn of premium with $3.3bn capitalised by the firm.
Assuming that Harvey losses had eroded premium, and that Irma and Maria wiped out the bulk, if not all, of two further pillars - say another $700mn for each - this would suggest losses of above 40 percent of capital.
Markel Catco marketing material seen by Trading Risk shows how three worst-case losses to the fund could produce a more serious outcome in a theoretical framework.
It outlined that a worst-case Florida hurricane, European windstorm and Japanese earthquake could result in a 45 percent gross capital loss.
This assumes that 30 percent of capital is lost to a Florida hurricane, 13 percent to a European windstorm and 25 percent to a Japanese quake - with 22 percent of a capital cushion from premium income that can absorb some of these losses.
The firm's pillared product does provide indemnity cover, but the attachment levels are designed to trigger in line with certain industry loss levels.
For example, a hypothetical deal structure outlined in a marketing presentation seen by Trading Risk lists a $30bn trigger for a Florida windstorm pillar.
Likewise, the Gulf wind trigger was listed at $15bn and the "rest of world" cover at $10bn - thresholds that would be put in play under Harvey and Maria based on current modelled loss estimates for the storms.
The initial loss estimate for Irma from loss reporting agency PCS was under $30bn, although models put it above this threshold.
In its annual report for 2016, Markel Catco said that 39 percent of its portfolio was exposed to North American/Caribbean risks.
In its statement this week, Markel Catco said that it planned to close its current fundraising round by the end of November, and noted that significant uncertainty surrounded the market impact of recent hurricanes affecting the US and Caribbean.
Markel Catco did not respond to repeated requests for comment.