The Underwriting Improvement Is Real, and a 17.7% Two-Day Re-Rating Has Already Spent It

Results vs. Consensus

Q3 2026 Scorecard

Two caveats on the top two rows before anyone quotes the percentages. The revenue consensus is not agreed: two providers put it at $338.9M, which makes the beat 4.5%, while a third implies $345.1M, which makes it 2.7%. And the "+900%" earnings surprise against a $0.01 estimate is an artifact, not a signal. Both the GAAP and the adjusted per-share lines for 2026 are dominated by $188.9M of nine-month Markel transitional amortization, and the GAAP line also by a non-controlling-interest split that routes most of the company's income away from Class A, so the Street's penny estimate carries almost no information. The lines that matter this quarter are Adjusted EBITDA, the loss ratio and the guide.

Year-Over-Year Comparison

Every figure below traces to the results release: the Condensed Consolidated Statements of Operations and, for the three adjusted lines, the non-GAAP reconciliation. The prior-year revenue comparatives are the Article 7 recast presentation the company now uses, which includes net investment income and gains in revenue. Under the superseded Article 5 presentation, Q3 2025 total revenue was $380M rather than $391.4M, so any comparison drawn against the older figure will understate the revenue decline.

The pair to sit with is Adjusted EBITDA up 63.5% and Adjusted Net Income up 6.2%. The gap is the whole 2026 story in two lines. Adjusted EBITDA excludes the $35.8M of Markel transitional amortization and adds back depreciation, share-based compensation and the tax line; Adjusted Net Income does not exclude the amortization, and it carries the tax. One of those numbers describes the operating business under its new structure and one describes what the owners actually earned this quarter, and the distance between them is why the per-share lines look nothing like the EBITDA line.

Quarter-Over-Quarter Comparison

The written premium and new-business declines are seasonal and expected: the second and third quarters are the driving season, the second is the peak, and the fourth is quiet. The two sequential moves worth noting are the loss ratio, down 1,010 basis points on the pick revision, and Marketplace revenue, down 18.9% in the quarter that contains Monterey Car Week, which is the segment's single largest event of the year.

Quality of Beat

Revenue. The 4.5% revenue beat is not an operating event. Total revenue fell 9.5% because the Markel Fronting Arrangement eliminates $137.9M of intercompany commission revenue in consolidation, and the quarter beat a consensus that was trying to model that elimination from the outside. The underlying revenue growth is in earned premium, up 41.7% to $265.1M, of which part is written premium growth and part is the quota share rising to 100% from 80% on January 1. Net assumed premium reached 79.4% of total written premium against 65.2% a year ago, which is the step-up showing up where it should. Set against that, the three fee lines were flat to down: Marketplace fell 6.4%, membership and other rose 1.9%, and commission and fee revenue is not comparable to anything.

Margins. The combined ratio of 79.7% is roughly ten points better than the 89.6% printed in each of Q2 2026, Q2 2025 and Q3 2025, and it is almost entirely the pick revision. Add the $26.0M back and the Q3 loss ratio is 42.4% against 42.0%, and the combined ratio is roughly 89.5% against 89.6%. On an annual basis, which is the honest basis for a company that books current-year losses off an annual loss ratio reviewed through the year, the nine-month accident-year loss ratio excluding the $5.7M of prior-year development is 38.5%, against 42.1% for the same nine months of 2025 and 40.2% after removing that period's $10.3M of Southern California wildfire losses. So the real underwriting improvement is on the order of 165 basis points on a catastrophe-adjusted basis, and it is real. It is just not 990 basis points.

EPS. GAAP diluted EPS of $0.08 fell from $0.11 and Adjusted Diluted EPS of $0.10 rose from $0.09, on a quarter when Adjusted EBITDA rose 63.5%. Three things sit in between. The $35.8M of transitional amortization is a real GAAP charge. The tax line swung $38.0M, from a $32.8M benefit on a valuation-allowance release last year to a $5.1M expense now. And the diluted share count collapsed from 347.2M to 107.8M because the prior-year calculation included the exchangeable units and this year's does not, which makes the reported diluted EPS series unusable across the two periods. The adjusted series, struck on a constant fully exchanged base of about 361.2M shares, is the only per-share line that compares.

Segment Performance

Insurance

Insurance segment pre-tax income of $48.9M was down 3.8% on revenue down 9.8%, which is a better outcome than those two numbers suggest once the moving parts are named. The segment absorbed the full $35.8M of transitional amortization and the loss of $137.9M of eliminated commission revenue, and still landed within $2M of last year on the back of $78.1M more earned premium and the $26.0M pick revision. Margin rose 100 basis points to 15.2%.

The operating detail underneath is genuinely strong in the places the thesis cares about. Policies in force rose 18.1% to 1,909,254. Nine-month new business count rose 56.1% to 402,168, which management calls a record for any comparable period. The State Farm Classic+ rollout moved from 37 states selling new business at the end of June to 40 at September 30, and 41 as of the call. The conversion of the 525,000 existing collector policies moved from 14 states in process at June 30 to 26 at September 30, a twelve-state step in a single quarter.

The most useful State Farm figure this quarter is in the 10-Q rather than the release or the call. Commission revenue generated from State Farm Classic+ policies was $8.1M in the quarter against $4.1M a year ago, and $21.6M for the nine months against $7.1M. That is a tripling year to date, and it is 35% of all consolidated commission and fee revenue for the period, which puts a number behind the CFO's comment that the line "is ramping quickly as State Farm conversions progress."

"The State Farm Classic+ rollout is progressing well, with our captive agents now selling Hagerty-powered policies in 41 states. The conversion of State Farm's existing Classic book of 525,000 policies is tracking to plan, with 26 states in process at quarter-end."
— McKeel Hagerty, CEO and Chairman

Assessment: the distribution engine is doing what the bull case needs it to do, the conversion pipeline accelerated materially, and the commission figure is useful. It is still only one line. Across three quarters of our coverage there has been no disclosure of State Farm written premium, policy count or margin contribution, so an investor can size the fee stream and still cannot size the underwriting one, which is where most of the economics sit.

Marketplace

Marketplace is the segment where the record headlines and the reported numbers point in opposite directions, and the reconciliation is in the revenue recognition. The Quail Auction during Monterey Car Week produced $87.6M of net auction sales, up $40.0M or 84.1% against the comparable prior-year auction. Aggregate auction sales rose 65.3% to $111.6M and private sales volume rose 88.1% to $99.9M. Marketplace revenue still fell 6.4%, because revenue from inventory sales dropped $9.6M, and when Broad Arrow sells a car it owns, the full sale price is revenue and the cost of the car is expense, whereas a brokered sale produces only the commission.

"When we sell a car from our own inventory, we record the full sale price as revenue and the cost of the car as expense rather than just a commission, so revenue can swing more based on the level of inventory sales."
— Patrick McClymont, CFO

The profitability direction is the better signal: the segment loss narrowed to $2.6M from $5.6M, a 54.1% improvement, with nine-month losses of just $0.7M against $6.0M. Finance revenue is part of it, driven by a 59.2% increase in the average Broad Arrow Capital loan portfolio to $154.6M, with a period-end balance of $162.3M against $103M of back leverage.

Assessment: Marketplace is close to breakeven on a transaction base that has roughly doubled, which is the right trajectory, and management's own caution about 2027 is the thing to carry forward. Values at the high end are up two to three times in a short period, management said it "would not be surprised to see prices take a breather," and the planning stance is explicitly not to assume continuation. A segment that is near breakeven at a cyclical peak is not yet a profit engine. The lending book is the part we would like disclosed before it doubles again: $162.3M against illiquid collateral with leverage behind it, and no loan-to-value, vintage or delinquency disclosure in the 10-Q.

Key Operating Metrics

Assessment: the net policy addition line is the one new piece of negative information in this print, and it is derived rather than disclosed. Policies in force rose 53,605 in the quarter against 57,433 in the same quarter last year, while gross new business rose 7.7%. The difference implies policies leaving the book grew roughly 22% against a book that grew 18.1%, which is the arithmetic consequence of retention running 40 basis points below last year on a much larger base. Retention itself stopped falling sequentially, which matters, but the aggregate churn is now large enough to be visible in the headline count.

Key Topics & Management Commentary

Overall Management Tone: confident and notably more specific than the previous two calls, with the quarter's single largest profit item disclosed, quantified and attributed without being asked. The posture on 2027 was the opposite: every forward question, on loss picks, on margins, on Marketplace run-rate, was deferred to the fourth-quarter call with the planning process cited, and the deferrals were consistent rather than evasive. The least convincing passage was the answer on net policy additions, where the explanation ran to seasonality and partner scheduling and did not engage the churn arithmetic the question implied.

1. The Loss Pick Came Down by About 3.5 Points, and That Is the Quarter

The company records current-year losses using an annual loss ratio set by its internal and external actuaries and reviewed through the year. In the third quarter it reduced that annual assumption, and the nine-month catch-up landed in one quarter as a $26.0M reduction to current-accident-year losses and loss adjustment expenses. A further $5.7M of favorable prior-year development, recorded in the first quarter against the 2025 accident year, takes the disclosed total to the $32M management quotes.

"Hagerty Re's combined ratio came in at 85.2% during the first nine months, helped by $32 million of lower estimated losses. This includes a $26 million reduction related to our current accident year recorded in the third quarter. This reflects lower physical damage costs and liability claims running below our estimates, and is in addition to the $6 million favorable development of 2025 accident year losses that we recorded in the first quarter."
— Patrick McClymont, CFO

The size of the revision is recoverable. Hagerty Re's nine-month losses were $286.0M on $756.7M of earned premium. Adding back the $5.7M of prior-year development gives a 2026 accident-year ratio of 38.5%. Adding back the $26.0M as well gives 42.0%, which is what the original pick must have been. So management lowered its own full-year 2026 loss assumption from roughly 42.0% to roughly 38.5% on nine months of claims experience.

Assessment: this is the most important disclosure of the year for this stock, and it cuts both ways. A 3.5-point reduction in the annual pick is a durable change in the earnings base, not a one-quarter windfall, and on roughly $1.18B of 2027 earned premium each point is worth about $12M. That is why our 2027 estimate goes up. It is also a revision made at the nine-month mark, which means the first two quarters of 2026 were over-reserved, and a pick that moved 3.5 points down on three quarters of data can move back.

2. The Underwriting Edge Pillar Finally Has Evidence Behind It

In August we marked the underwriting-edge pillar at risk rather than on track, because the first-half loss ratio improvement was entirely a first-quarter event that reversed: 38.4% in Q1 2026 became 42.7% in Q2, against 42.3% in Q2 2025. Three quarters of 2026 data now say something different. Excluding reserve items from both periods, the 2026 accident year is running at 38.5% against 40.2% for the same nine months of 2025 after removing that period's wildfire losses.

Management attributes the improvement to three things, two internal and one not. The claims organization is roughly 40% larger than two years ago, with the material damage and special investigation units doubled. Liability frequency is down across the industry post-pandemic. And the modest rate increases the company has taken were targeted at liability lines specifically.

"High drive-away rates are a big reason our losses have consistently come in better than daily driver CAT models would predict, and why reinsurers and AM Best value this lack of correlation with daily driver losses. Our loss ratios are much more stable than the industry's, with roughly 35% lower volatility since 2010, which is also a compelling story for reinsurers as we formalize next year's program at attractive prices."
— Patrick McClymont, CFO

Management also quantified the catastrophe sensitivity in a useful way. Hurricanes Helene and Milton cost $27M in 2024 against $162M of Adjusted EBITDA; at current profit levels, equivalent events would have roughly half that proportional impact. The company has had two quiet catastrophe seasons in a row.

Assessment: we are moving this pillar to on track. The caveat worth holding is that the industry frequency tailwind is not proprietary and the two quiet catastrophe seasons are luck, so the claims investment is the part of the improvement that is a durable competitive advantage. With the quota share at 100%, every basis point of loss ratio now costs 25% more than it did in 2025, so this metric remains the one that would break the thesis fastest in either direction.

3. The Acquisition-Cost Deferral Is Finished, and Management Told the Street to Strip It Out

Under the fronting arrangement the MGA subsidiary's policy acquisition costs are deferred and amortized over the policy term, where in 2025 they were expensed as incurred. Because the deferred asset was built from zero this year, the deferral ran far ahead of the amortization, and the gap is a pure Adjusted EBITDA benefit. Nine-month figures: $88.1M deferred against $36.4M amortized, for a $51.7M benefit. The third quarter contributed $30.8M against $20.1M, or $10.7M.

"This outlook includes the $52 million benefit from capitalizing MGA acquisition costs in 2026, a timing effect that is largely complete as of September 30th and will not repeat again in 2027."
— Patrick McClymont, CFO

This is a clear improvement in disclosure. Last quarter the $41M first-half benefit and the $153M of transitional cost were explained separately and never put on the same page, which was our central criticism of the Q2 presentation. This quarter the benefit was quantified in prepared remarks, flagged as non-repeating, and then, when asked, management told the Street exactly how to treat it in a 2027 bridge.

Assessment: credit where it is due, and the arithmetic is now unavoidable. The guided 2026 Adjusted EBITDA midpoint of $310M contains $52M that will not exist next year. Anyone growing $310M at mid-teens to frame 2027 is building off a base $52M, or 20%, above the one management just told the Street to use.

4. Remove the Deferral and the Margin Expansion Disappears

Management said on the call that 2026 will deliver margin expansion, as guided. On the reported basis that is unambiguously true: Adjusted EBITDA of $310M at the guide midpoint against $1,391M of written premium is 22.3%, against 19.8% in 2025. Remove the $52M deferral benefit from 2026, adjust nothing in either year's reserves, and the same ratio is 18.5% against 19.8%, a 130 basis point contraction.

"If you look at the numbers, we have continued to expand margins since we began that journey sort of in 2023. We will see how this year wraps up, but we will expect that we are expanding margins again this year, which is what we guided to."
— Patrick McClymont, CFO

This matters more than it would at most insurers because 2026 is the year the economics permanently improved. The quota share went to 100% from 80% and the ceding commission fell from 47% to a roughly 2% fronting fee plus self-funded expenses, which management describes as a 25% step-up in retained underwriting profit and investment income. A step-up that large should move the profit-to-premium ratio up, not down.

Assessment: the honest reading is that the economics step-up is real and has been spent on growth. Management says as much, repeatedly and without embarrassment, and the spend is going to defensible places: claims capacity, Duck Creek, the member service platform, agent enablement. It is still a fact that the first year of permanently better economics shows no margin gain on a deferral-adjusted basis, and it is the fact that most undercuts the case for paying a higher multiple for 2027.

5. The Implied Fourth Quarter Is the First Clean Quarter, and It Is Guided Up Sharply

Last quarter we argued the raised guide implied a flat second half, and that call was wrong. We said the $270M to $280M range left $110M to $120M for the second half against $116.0M a year earlier. The third quarter alone delivered $97.1M, and the new range leaves $48.3M to $58.3M for the fourth, so the second half is now tracking to $145.3M to $155.3M, up 25% to 34%. Most of the difference is the pick revision nobody could have forecast; about $9M is operating upside.

The fourth quarter carries no deferral benefit by management's own account and only about $10M of residual transitional amortization. Against a Q4 2025 that contained a $21M reserve reduction, the implied range represents something like 36% to 63% underlying growth.

Assessment: this is the most encouraging line in the guide and the one the market is right to focus on. The caution is that a flat-to-down headline Q4 against a flattered prior year is exactly the kind of optic that produces a disappointing-looking print in February, and that the implied Q4 revenue of $319M to $333M sits below the roughly $348M the Street carries. Revenue estimates have to come down while profit estimates go up.

6. Written Premium Decelerated to 15.5%, and Net Policy Adds Went Backwards

Written premium growth of 15.5% in the quarter compares with 19.2% in the second quarter and 18.3% in the first. Management framed it as a two-year stack against harder comparisons from the second half of 2025, which is fair. The harder number is the policy count. Policies in force rose 53,605 in the quarter against 57,433 in the third quarter of 2025, a 6.7% decline in net additions, in a quarter when the State Farm conversion was live in twelve more states than it had been in June.

Gross new business rose 7.7% to 123,321. Net additions fell. The difference implies policies leaving the book rose roughly 22% year over year, against a book 18.1% larger.

Assessment: this is the quarter's quietest negative. Growth is still high and still count-driven, which is the pillar, but the composition is shifting: more gross adds are being required to produce fewer net adds, and the conversion pipeline is doing more of the work. If the fourth quarter repeats the pattern with the conversion program in 26 states, the question stops being seasonal.

7. A Softening Personal-Lines Market Cuts Both Ways on the Partner Funnel

A meaningful share of new business arrives downstream of the large carrier partnerships, and the personal auto and home market is softening: rates are moderating and availability is improving in exactly the states where constrained markets had been pushing business toward specialty writers.

"A lot of the big carriers are expressing softening new business growth. We tend to be positioned really well in the sense that, again, retention is very strong. That is a big driver of how the book of business looks, and especially given the fact that a very large amount of our new business is attributed to high Net Promoter Score customers referring new business to us."
— McKeel Hagerty, CEO and Chairman

Assessment: the answer is reasonable and it is not a forecast. If partner new-business flow is softening and Hagerty's own net additions are already running below last year, the two facts point the same direction into 2027. The offset is genuine: a hardening market a few years ago was a tailwind, and nobody marked the stock down for it then.

8. Collector Car Values Are at a Peak, and Management Said So

The cycle commentary was unusually candid for a company whose marketplace segment benefits from rising values. Total Monterey sales across all auction houses reached a record $756M in August, up 75% year over year, with modern supercars and hypercars accounting for half. Broad Arrow set nine price records at The Quail, Zoute in Belgium rose 35% in October, and the Wynn Las Vegas sale was seven times the prior year's inaugural event.

"I would not be surprised to see prices take a breather at some point after the sharp rise in 2026, but I believe the long-term trends on these increasingly scarce assets remains up and to the right."
— McKeel Hagerty, CEO and Chairman

On the planning question the CFO was more explicit still, saying that for 2027 "the prudent thing to do is not to assume that that will continue as is" and that the company does not "want to plan for trees growing to the sky."

Assessment: management is pre-announcing a conservative 2027 Marketplace plan, which is the right call and a modest negative for anyone extrapolating the auction line. The second-order effect is the one to watch: rising values support the insurance book's guaranteed-value pricing and the lending book's collateral, so a genuine cooling touches more than one line.

9. The Claims and Technology Spend Is Showing Up in Unit Costs

Two disclosures gave the operating leverage argument something measurable behind it. The member service centre now requires one full-time employee for every 2,800 policies in force, a 20% year-over-year productivity improvement across a 1.9 million policy book, attributed to artificial-intelligence-assisted call summaries and workflow automation. Separately, the special investigation unit is described as having produced millions of dollars of benefit.

The member service disclosure came in prepared remarks and is the most concrete unit-cost figure the company has given.

"Our member service center is a great example, where we now require just one full-time employee for every 2,800 policies in force. This is a 20% year-over-year productivity improvement across a book of 1.9 million policies. AI-assisted call summaries and workflow automation are handling tasks that previously required manual effort."
— Patrick McClymont, CFO

Assessment: a 20% unit-cost improvement on the largest labour pool in the business is a real number and the kind of disclosure we have been asking for. It is also $9M of a $35M raise, which is the honest scale of the operating upside in 2026, and it is why we credit roughly two points of margin expansion in 2027 rather than more.

10. The 48% Operating Cash Flow Increase Is Mostly Balance Sheet Migration

Nine-month operating cash flow of $280.1M was up 47.5% from $189.9M, and management presented it as evidence of the improved economics. Most of it is not earnings. Net income plus non-cash adjustments produced $41.8M of operating cash flow before working capital, against $165.0M in the prior year. The entire increase comes from a $238.3M working capital inflow against $24.9M last year, of which unearned premium growth contributed $230.7M, the release of the old deferred acquisition cost balance contributed $103.7M, and receivables contributed $118.3M, offset by $75.6M of amounts due to insurers, $81.6M of ceding commissions payable and $32.1M of loss reserves.

Assessment: the float build is real cash and it is genuinely better than the old structure, because premium now flows to Hagerty Re rather than through Markel. It is a one-time migration of working capital onto the balance sheet plus the ordinary float growth of a growing book, not a 48% improvement in cash earnings. The pre-working-capital figure of $41.8M is the one to carry.

11. Capital: $461M of Cash, a Bigger Loan Book, and Buybacks Still Last

Unrestricted cash reached $460.8M against $160.2M at year end, with total debt of $246.4M including $103M of Broad Arrow back leverage. The cash balance drew a question and the answer was mechanical: the company is shifting the investment portfolio modestly down the credit curve and was mid-execution on two transactions at quarter end. Equity securities rose to $84.1M from $34.9M over the nine months while fixed maturities fell.

The Bennetts acquisition closed on October 8 for £34.5M, approximately $46.0M in cash, and will be reported in the Insurance segment. The 10-Q calls it not material. The capital allocation ranking management set out last quarter is unchanged: organic investment first, bolt-on acquisitions second, return of capital third and explicitly not on the near-term agenda.

Assessment: the balance sheet is in good shape and the priority ordering is correct for a business compounding policy count at 18%. The two items we would still like disclosed are the Broad Arrow loan book's credit characteristics, now $162.3M and growing 62% against illiquid collateral, and the risk appetite of the investment portfolio shift, which is being described qualitatively while the equity allocation more than doubled.

Guidance & Outlook

This is the second consecutive quarter of a material raise. In August the written premium range rose with the profit ranges. This time it was left alone while the profit ranges moved. Management now expects written premium growth at the high end of the 16% to 17% range, which implies $1,397M and a fourth quarter of roughly $297.6M, up 14.8%. The revenue raise simply reflects a smaller eliminated commission base than previously modelled; revenue is still guided down 8% to 7% for the year.

The Markel transitional amortization path is tracking as guided: $89M in Q1, $64M in Q2, $36M in Q3 and approximately $10M expected in Q4, for roughly $199M in total, after which it is gone. That is the single largest reason the 2027 GAAP presentation will look unrecognizably better than 2026, and it is also the reason GAAP net income guidance of $48M to $60M says nothing useful about earning power.

Implied Q4 ramp: Adjusted EBITDA of $48.3M to $58.3M against $256.7M booked, with no deferral benefit and about $10M of residual transitional cost. Against the $56.6M reported in Q4 2025, the midpoint is down 6.0%; against the roughly $35.6M that quarter produced excluding its $21M reserve reduction, the midpoint is up about 49%.

Street at: the raised revenue range brackets a consensus cited at roughly $1.34B and sits below a figure cited elsewhere at $1.36B, so the guide is in line to slightly light on revenue depending on which print you use. The implied Q4 revenue of $319M to $333M is clearly below the roughly $348M carried for the quarter, so revenue estimates must come down. On profit there is no reliable published Adjusted EBITDA consensus, but a $35M raise with one quarter left is unambiguously above where the Street sat.

Guidance style: management has now beaten or met expectations for fifteen consecutive quarters by its own count, and has raised the profit guide twice this year. The pattern is conservative initial framing followed by upward revision, and the written premium range has moved far less than the profit ranges. The one thing conspicuously absent from a company with this guidance record is any 2027 framework at all, which is the subject of the next section.

Analyst Q&A Highlights

Why Net Policy Additions Fell While the Conversion Programme Accelerated

The call opened on the softest number in the release, which is not disclosed directly and has to be derived from the policy count. The question noted that sequential net policy additions were lower than a year ago despite the State Farm conversion being expected to contribute more, and asked whether retention was the cause. The answer separated the two: retention is where it was planned, and the shortfall is seasonality and the timing of state-by-state rollouts. Neither half of the answer engaged the implication that gross additions grew 7.7% while net additions fell 6.7%.

Q: "Sorry to nitpick here a little bit, but if I look at the quarter-over-quarter net policy count growth, the net new PIF is slightly lower year-over-year, which is a little surprising, just given the State Farm conversions I would think would be more meaningful this year. It looks like it's maybe retention-driven. Just wondering if you could update us on what you're seeing there."
— Charlie Lederer, BMO Capital Markets

A: "No, retention continues to be consistent with historical levels and right where we had planned it for, so there's no concerns on the retention front. On the PIF growth front, yeah, some of this is the seasonality, right? It's always going to be a little bit trickier to predict as we get deeper into the year, and some of it is the timing of State Farm. So we continue to roll out states, we continue to move existing states into conversion, but that can move around a bit depending on their schedule and rollouts, and so it can move. There's nothing that's giving us pause on that front."
— Patrick McClymont, CFO

Assessment: the most revealing exchange of the call, because the questioner found the number without the company presenting it and the response did not address the mechanism. Retention being "right where we had planned it" is consistent with 88.2% against 88.6%, so planned deterioration is still deterioration. The fourth quarter, with the conversion live in 26 states, will settle whether the explanation was seasonality.

What Actually Drove the $35M Guidance Raise

The most useful answer of the call came from the last analyst on it. A questioner asked directly whether the full-year raise was entirely the better loss experience or whether operating expense and revenue contributed, and management decomposed it on the spot. That single answer does more for the integrity of the 2026 numbers than anything in the prepared remarks, and it is the reason this report can state the quality of the raise rather than estimate it.

Q: "When we look at the full-year adjusted EBITDA guide, so it increased by about $35 million at the midpoint. In your guys' own internal projections, was the improvement this quarter entirely driven by the better loss and claims experience in the loss ratio, or was there anything else on the operating expense side or revenue side that allowed you to raise the full year guide?"
— Tommy McJoynt, Keefe, Bruyette & Woods

A: "You talked about the $35 million is the increase, right? $26 million of that is what we released in this quarter. The other $9 million are strong operations within the business. The way that we think about that is a fair bit of that is the fact that there is operating leverage inherent in the business. We are being disciplined, and we are making investments, we talked about earlier on the call, but we are also being disciplined in terms of our cost structure and our headcount. That leverage is working. It is not entirely the reserve releases. There are other benefits in the business that are flowing through."
— Patrick McClymont, CFO

Assessment: volunteered dollar precision, when asked, on the least flattering decomposition available, which is the opposite of the Q2 presentation and should be noted as such. The content is the content: 74% of the raise is a reserve item and 26% is operations. A $9M operating beat on an implied second half of roughly $115M is about 8%, which is a good quarter and not a re-rating event.

Whether the $52M Deferral Benefit Is a Straight 2027 Headwind

The 2027 bridge question was asked plainly and answered plainly, which has not always been the case on this topic. The questioner asked whether the capitalization benefit should simply be treated as a headwind when the 2027 guide arrives, and management confirmed that it should and explained that this is why the item has been quantified on every call this year.

Q: "So you guys have, there was a $52 million benefit, right, this year from capitalizing, right, some of those acquisition costs, and that goes away next year. When we think about adjusted EBITDA, do we just kind of think about this $52 million somewhat being a headwind, right, relative when you put in place next year's guidance?"
— Elyse Greenspan, Wells Fargo

A: "On the first one, yeah, I think as you think about bridging 2026 to 2027, you should kind of start your 2026 by adjusting that out or treating it as a sort of a one-time item that was a tailwind this year that goes away next year. That is the right way to think about that, and that's why we've been so clear about it all year long to make sure people understand that dynamic."
— Patrick McClymont, CFO

Assessment: the answer we wanted and did not get last quarter. The clean 2026 base is now a matter of company guidance rather than analyst construction, which removes the disclosure half of our first bear point while leaving the arithmetic half entirely intact. A reader who takes management at its word is modelling 2027 off roughly $258M, not $310M.

Whether Lower Loss Picks Carry Into 2027

The natural follow-on to a 3.5-point reduction in the annual loss assumption is whether next year gets priced off the new number, and a questioner asked exactly that. The answer was a deferral, delivered lightly, and it leaves the single largest swing factor in the 2027 estimate unresolved until February.

Q: "With the better loss experience, sounds like you are doing really well with the internal claims capability. Is there a potential for loss picks potentially to be a little bit lower?"
— Mark Hughes, Truist Financial

A: "Yeah. We will tell you when we tell you, Mark. We are going through that process now, and we will share guidance on the fourth quarter call. We are pleased with both sort of our internal performance and then obviously the market trends as we have talked about. So, that is part of our consideration set as we do the plan for next year, but nothing to preview right now."
— Patrick McClymont, CFO

Assessment: a legitimate deferral that happens to sit on the most valuable unknown in the model. Each point of 2027 loss ratio is worth roughly $12M of Adjusted EBITDA on our earned premium estimate, so the spread between a 38.5% pick and a 41% pick is about $30M, or roughly 10% of our 2027 number. The deferral is also a reason the February call is a two-sided catalyst rather than a one-sided risk.

Whether the Liability Improvement Is Hagerty or the Industry

A follow-up pressed on the source of the liability improvement, which is the component of the loss ratio least within the company's control. The answer separated what the company owns from what it borrows, and was more useful for being candid about the split.

Q: "When you think about the improved performance on the liability claims, you just alluded to the fact that you're seeing broader market improvement. I am not sure if it is the liability specifically, but anything you can say about that? Seems significant. Just curious what your experience might be telling you on the liability front."
— Mark Hughes, Truist Financial

A: "So, we have been boosting our capabilities on Material Damage, which is how we internally adjust for losses and work with repair shops, restoration shops to get our members' cars back on the road. The SIU, which is, of course, Special Investigative stuff, is really meant to make sure that we are managing our results by carefully looking at losses to make sure that there is no fraud involved with them. You cannot price for fraud in insurance, so you have to be really careful when you identify instances of it. We are really proud of the teams really having millions of dollars of benefit to the business as we are being very, very careful with it. On the liability side, frequency is down. So, it is a unique thing that is happening right now post-COVID when there was a lot of people out driving, a lot of people buying all sorts of cars. Frequency is down, and so that does have a downstream impact, of course, in liability losses."
— McKeel Hagerty, CEO and Chairman

Assessment: the honest answer is that the physical damage and fraud improvements are company-specific and the liability improvement is largely industry frequency. That split matters for durability. A frequency cycle that reverses takes back the borrowed half, and the company has no more control over that than any other personal lines writer.

How Much of the Marketplace Result Should Be Run-Rated

Management was asked how much of the year's auction and private sale strength an investor should extrapolate, and chose to answer with cycle caution rather than a number. The reply is the clearest statement yet of how the 2027 Marketplace plan will be framed.

Q: "Maybe just one last one, just in the marketplace business, just on your comments on the surge in prices at the top end, McKeel, potentially taking a breather. Just wondering if you guys have a perspective or view on how much of this quarter's or year-to-date marketplace revenues we should think about run rating or not think about run rating going forward."
— Charlie Lederer, BMO Capital Markets

A: "I think we just recognize that for those type of cars, the valuations are up 2x, in some cases 3x, in a relatively short period of time. It would be reasonable to expect that at some point that flattens, and potentially you could even see that cooling down a bit. That is what we are communicating there. The overall market continues to be pretty stable for collector cars. We are going through our planning process right now for next year, and so we have not finalized that yet. What I will say is it is similar to how we talk about private sales. You are talking about a relatively limited number of cars driving some pretty big outcomes in the market and for us specifically. As we look forward to 2027, the prudent thing to do is not to assume that that will continue as is, certainly on the kind of the upward slope that it is on."
— Patrick McClymont, CFO

Assessment: pre-managing the 2027 Marketplace guide downward, which is sensible and mildly negative for the segment line. The more interesting admission is that a small number of cars drives the outcome, which is the same concentration risk that made nine-month private sales fall 18.4% while the third quarter alone rose 88.1%. This is a volatile revenue line being presented as a flywheel.

Whether Investment Spending Keeps Capping Margin Expansion

The structural version of the margin question got asked: as the business scales toward its 2030 target, does the investment burden ever let earnings catch up, or does each new opportunity absorb the leverage. The answer was a framework rather than a number, and it conceded more than it resolved.

Q: "You have been clear in the past about talking about investing well ahead of growth, but those costs probably would not go away, I guess, as the business grows. I want to get a sense of, as we look past maybe 2027 and into the longer term, 2028 and prior, and after that, are we approaching a period where the earnings grow, where those investments kind of flow through back into earnings, or do more investment opportunities keep that spending growth kind of elevated?"
— Michael Phillips, Oppenheimer

A: "It is a balancing act, right? We want to continue to march on a path to what we think is the right long-term margin structure for the business. But because we are deep enough into that path, we do have the flexibility to invest well ahead of growth. That is a balancing act that our leadership team and McKeel and myself, we spend a lot of time on. We are going through it right now for the planning process. If we want to keep growing for the long haul, not just double by 2030, but continue to have this be a growth business, there always will be opportunities to invest to support that growth. Our job is to strike that balance and kind of give you communications on a year-by-year basis how we think the margins are trending."
— Patrick McClymont, CFO

Assessment: a candid answer whose content is that the margin path is a management choice made annually rather than a structural trajectory an investor can underwrite. That is exactly what the deferral-adjusted margin contraction in 2026 shows, and it argues for valuing this business on growth at a stable margin rather than on growth plus margin expansion.

What They're NOT Saying

1. The revised full-year 2026 loss-ratio pick. The $26.0M reduction implies the annual assumption moved from roughly 42.0% to roughly 38.5%, and that figure has to be reverse-engineered from the reserve reconciliation and the earned premium base. The number management now carries is never stated.
2. The third quarter's own clean Adjusted EBITDA growth. Every growth framing on the call ran through the nine-month figure, where the pick revision is spread across three quarters of earned premium. The quarter's 63.5% increase is 97% explained by two disclosed items, and the quarter's clean growth rate was never put on a slide.
3. Why net policy additions fell year over year. Asked directly and answered with seasonality and partner scheduling. The lapse or cancellation count that would settle it is not disclosed in any period, so the 22% implied increase in policies leaving the book has to be derived.
4. State Farm's written premium, policy count and margin contribution. The fee line is sized, at $8.1M for the quarter and $21.6M year to date, and only in the 10-Q. The underwriting economics are not. The state counts are granular to the unit, 40 selling and 26 converting at quarter end, while the premium and the margin behind them remain absent across all three quarters of our coverage.
5. Enthusiast+ anything. Live in Colorado plus three additional states since July. No loss ratio, no policy count, no premium, and on price sensitivity, "it is too early to draw any conclusions." This was the product whose pricing assumptions were revised once already.
6. That the operating cash flow increase is working capital. Nine-month operating cash flow of $280.1M was presented as evidence of the new structure's economics. Before working capital it was $41.8M against $165.0M a year ago. The $238.3M working capital inflow, of which $230.7M is unearned premium and $103.7M the release of the old deferred cost balance, was not characterized.
7. The loss portfolio transfer gain. Hagerty Re assumed the pre-2026 Essentia liabilities for $50.5M of cash, carries an estimated $39.2M of ultimate liability and an $11.3M deferred gain, and has recognized $4.5M of that gain year to date inside interest expense and other, net. It is why the Insurance segment's interest line is a credit this quarter. Not mentioned on the call.
8. Broad Arrow Capital's credit book, again. $162.3M at quarter end, up 62.2%, with $103M of back leverage against collector-car collateral. The 10-Q says borrowers must maintain "a specified targeted loan-to-value" ratio to renew and never says what that ratio is, and there is no vintage or delinquency disclosure anywhere in the filing. The only question in the area was about the cash balance.
9. The 2027 revenue base. On the May call management said it expects 2027 revenue growth to track written premium growth more closely, with written premium growing mid-teens. Off what base is unstated. Nine-month consolidated commission and fee revenue is $61.7M against $447.0M on the MGA+ standalone basis, and the gap is a consolidation elimination whose 2027 path has never been framed.
10. Bennetts economics. £34.5M, approximately $46.0M in cash, closed October 8, described as not material and reported inside Insurance. No premium, policy count, combined ratio or expected contribution.
11. The investment portfolio's new risk appetite. Equity securities rose to $84.1M from $34.9M over nine months and management says it is "modestly moving into riskier assets," mainly credit investments "further down the credit curve." No target allocation, no credit quality, no duration.

Market Reaction

• Pre-print setup: HGTY closed at $15.89 on November 2, up 18.2% year to date against an S&P 500 up 10.2%, up 16.8% over the prior 30 days and up 40.7% over the trailing twelve months. The 52-week closing range entering the print was $9.67 to $15.95, so the stock was sitting at the very top of its own range. That is the mirror image of the Q2 print, which was taken from the bottom half of the range.
• Reaction session: the release came before the open on November 3. The stock gapped to $16.58, traded $16.11 to $17.29 and closed at $17.03, up 7.2% or $1.14. The S&P 500 rose 0.4%.
• Follow-through session: November 4 opened at $17.29, traded $16.92 to $18.73 and closed at $18.71, up a further 9.9% or $1.68, a new 52-week closing high. The S&P 500 rose 1.6%.
• Volume: 0.6 million shares on the reaction day against a 0.4 million 30-day average, then 1.0 million on the follow-through day, roughly 2.9 times normal. The heavier session was the second one.
• Two-session cumulative: $15.89 to $18.71, up 17.7%, against an S&P 500 up 2.0% across the same two days.
• Sell-side response: every target change with a known prior target was an increase, and the increases were large: one to $19 from $15, one to $17 from $14, one to $16 from $13. No rating migrated. Two of those three firms remain at neutral, as does a fourth firm that set a $14 target. The highest standing target was reaffirmed rather than raised, and the post-revision average target sits below the second session's close.

The first day's move was the right reaction to what was legible in the first ten minutes: an earnings and revenue beat, a $35M profit guidance raise, a combined ratio with a ten-point improvement on it, and a record nine-month count of new members. A stock at the top of its range with a neutral Street rarely gets a 7% day out of that, and it got one.

The second day is the informative half, and it inverts the Q2 pattern exactly. After the August print, the stock rose 15.4% on the day and then gave back about two-fifths of the move over the following two weeks on falling volume, as the market worked through an Adjusted EBITDA growth rate that did not survive a symmetric adjustment. This time the second session was larger than the first on 1.7 times the volume. The second read was better than the first, not worse, and we think the reason is specific: the first day traded the guidance raise, and the second day traded the loss pick. A 3.5-point reduction in a specialty insurer's annual loss assumption, at 100% quota share, is a larger and more durable event than a $35M guidance raise, and it takes a day of reading the reserve reconciliation to see it.

What the two sessions did not do is change anybody's mind about the stock. Five firms acted and none moved its rating. The composite picture is a Street that marked its models up sharply and its conviction not at all, which left the price above the revised average target. That is a crowded place for a stock to sit into a February call that has to put a number on the $52M headwind.

Street Perspective

Debate: Is the Loss Pick Revision a Re-Rating Event or a One-Quarter Optic?

Bull view: the company books current-year losses off an annual actuarial loss ratio and it just lowered that ratio by about 3.5 points on nine months of data, with physical damage, fraud recovery and liability frequency all contributing. At 100% quota share on roughly $1.18B of 2027 earned premium, each point is worth about $12M. This is a permanent improvement in the earnings base that the market is only beginning to capitalize, and the fact that 2027 may be priced off the new pick is upside nobody has in a model.

Bear view: it is a nine-month catch-up recognized in one quarter, which flatters the reported combined ratio by nearly ten points and makes the quarter look like something it was not. Industry liability frequency, one of the three stated drivers, is a cycle rather than an edge, and two quiet catastrophe seasons are weather. A pick that moved down 3.5 points on three quarters of data can move back on four, and the revision means the first half was over-reserved.

Our take: the bull has the better case and the bear has the better calendar. The improvement is real and we are raising our 2027 estimate because of it, from $270M to $295M. But the company itself will not commit to carrying the lower pick into next year, the industry frequency component is borrowed rather than owned, and the stock priced the full benefit inside two sessions. Being right about the earnings base and wrong about the entry price is still being wrong.

Debate: Can the 2027 Base Be Known Before the February Call?

Bull view: it can, because management has done the work for you. The $199M of transitional cost ends in December. The $52M deferral benefit is quantified and management has told the Street to strip it out. The quota share is permanent, written premium compounds mid-teens, operating leverage is now measurable in unit costs, and the conversion pipeline is contracted. Start at $258M, grow it mid-teens with margin, add the lower loss pick, and 2027 clears $300M on a clean GAAP presentation.

Bear view: the base is knowable and the growth rate is not. The deferral-adjusted margin on written premium went backwards in 2026, in the first year of a 25% economics step-up, because the step-up was spent. Management describes the margin path as an annual balancing act rather than a trajectory. Marketplace is being pre-guided down, net policy additions are falling, written premium is decelerating and the partner funnel is softening with the P&C cycle. A $258M base growing 8% to 9% lands near $280M.

Our take: our $295M sits between them and closer to the bull, which is a change from last quarter. The deferral criticism survives in full as arithmetic and has been resolved as a disclosure complaint, which is progress. The risk that matters is not the base, it is what the February guide does to a share price that has already capitalized the better base at 22.4 times.

Debate: Does Decelerating Policy Growth Matter While the Pipeline Accelerates?

Bull view: no, because the composition is improving where it counts. The conversion programme went from 14 states to 26 in a single quarter and new-business sales were live in 40 states at quarter end and 41 as of the call, nine-month new business is a record 402,168, and the 525,000 policy book is contracted through 2028. Quarterly net additions are a noisy number affected by state scheduling, and a book growing 18% with 88% retention and a Net Promoter Score roughly double the industry is not a decelerating franchise.

Bear view: gross additions grew 7.7% and net additions fell 6.7% in the quarter when the pipeline was supposed to be contributing most, which means attrition is rising faster than the book. Written premium growth has fallen from 19.2% to 15.5% in one quarter. Retention has been below the prior year for three consecutive quarters, vehicles per policy are down 4.0% and the membership attach rate is down 570 basis points, so each incremental policy is worth less than the one it replaces.

Our take: the bear is describing a real and compounding dilution and the bull is right that none of it threatens the medium-term policy count. The resolution is a disclosure the company will not give: if State Farm conversions were separable, an investor could see whether the core book is growing at all. Until then, a decelerating headline with rising churn is a reason to pay a fair multiple rather than a premium one, which is the same conclusion we reached in August by a different route.

Debate: Is the Collector Car Cycle a Tailwind or a Peak?

Bull view: the generational handover is structural, not cyclical. Gen X, Millennial and Gen Z buyers now account for over 60% of quote volume and they want 1980s to 2000s cars that are becoming scarce. Record Monterey sales of $756M across all houses, auction sales up 83% year to date, nine price records at The Quail and a new Barrett-Jackson partnership are evidence of share gain inside a growing market, and rising values support guaranteed-value pricing in the insurance book.

Bear view: top-end values are up two to three times in a short period, the CEO said he would not be surprised to see prices take a breather and the CFO said the company will not plan for continuation. Marketplace revenue fell 6.4% in the quarter containing Monterey because the gross-versus-net revenue mix swung, and nine-month private sales fell 18.4% on the loss of a single large collection. A near-breakeven segment at a cyclical peak, sitting on a $162.3M levered loan book against the same collateral, is a liability in a cooling market rather than an asset.

Our take: the demographic argument is sound and the segment is still too small to move the valuation. The lending book is not too small to matter in a downturn, and management has now said twice that it does not expect the current slope to persist. We model Marketplace flat to slightly down in 2027 and we would like loan-to-value and vintage disclosure before that portfolio doubles again.

Model Update

Aardvark Labs does not maintain a published financial model for Hagerty. The framework below is what the rating rests on, with each change from the August framework stated and sourced.

Capital structure. On a fully exchanged basis the share count is 361.2 million, which is the company's own adjusted diluted denominator and assumes conversion of the 241.0 million exchangeable units held by the non-controlling interest, the Series A preferred and unvested awards. At the November 4 close of $18.71 that is $6,759M of equity value. Adding $246M of debt and the $63M tax receivable agreement liability and deducting $461M of unrestricted cash gives an enterprise value of roughly $6,607M. Restricted cash of $133M is insurance float and is excluded.

The re-rating added three and a half turns to the forward multiple in two sessions. That is the whole case for the downgrade in one line: our estimate for the first clean year rose 9%, and the price rose 17.7%.

Scenario Framework and Target

12-month target: $16, from the base scenario, implying roughly 14.5% downside from the November 4 close. The bear case assumes the loss ratio gives back two points toward 40.5%, Marketplace cools and written premium growth slips to the low teens. The bull case assumes the 38.5% pick holds or improves, full operating leverage flows through, Marketplace holds its 2026 level and the market pays a franchise multiple for the 2030 policy target once a clean year is printed. Note what the bull case produces: 6.4% upside, which is roughly a market return. When the optimistic case pays you the index and the base case costs you nearly 15%, the distribution is doing the arguing.

The catalyst. The fourth-quarter call, expected in late February 2027, is the first time management will attach a number to 2027. Three of this call's most important questions were deferred to it: the 2027 loss pick, the 2027 margin path and the Marketplace plan. The guide will be built off a clean base near $258M by management's own instruction, and the Street is marking models up off a reported $310M. That gap is the risk, and it is a two-sided one: a 2027 Adjusted EBITDA guide starting with a three would vindicate the bulls and send us back to work.

Thesis Scorecard Post-Earnings

The pillars below are the ones carried in our standing thesis since we restated it in August, graded against what this quarter showed. One pillar is added this quarter and labelled as such.

Overall: the thesis is stronger on the business and weaker on the stock. Three of four bull pillars held and the fourth was upgraded on the first hard evidence we have seen for it, which is a better quarter than we expected when we set these pillars in August. Two of our criticisms of the Q2 presentation were answered directly: the deferral benefit was quantified in prepared remarks and the guidance raise was decomposed on request. Against that, the accounting arithmetic of the first bear point is unchanged, the per-policy dilution continues, churn is now visible in the headline count, and the share price moved 46% above our August anchor while our estimate for the first clean year moved 9%.

Action: Underperform, target $16. We would return to Hold on a pullback toward $16 without a deterioration in written premium growth, or on a 2027 Adjusted EBITDA guide at or above $300M built off a deferral-adjusted base. We would move to Outperform on a 2027 guide above $320M with the lower loss pick carried into the plan and the State Farm premium contribution disclosed in dollars. We would stay at Underperform or lower if the loss ratio gives back more than two points, if core retention breaks below 87%, or if a fourth quarter of falling net policy additions confirms that the third quarter was not seasonal.