ticker: ARX task: earnings-review period: Q3_FY25 quarter_end: Sep-25 date: 2026-02-22 analyst: Phil Fisher

Accelerant Holdings (ARX) — Q3 FY25 Earnings Review

Philip A. Fisher | February 22, 2026


Prior Beliefs Going In

This is my first formal analysis of Accelerant Holdings. The company IPO'd in mid-2025 and has produced only two quarters of public financial disclosures. I approached this without a prior thesis — which is itself instructive. I have spent considerable time with the available data, and I must say that what I have found here is one of the more genuinely interesting business structures I have examined in the insurance industry in many years. Whether it qualifies as an outstanding growth company in the Fisher sense requires careful work.


Scuttlebutt Assessment

The scuttlebutt on Accelerant Holdings is limited by the company's youth as a public entity and its niche positioning within specialty insurance. I pursued seven categories of inquiry.

1. Customers (Member MGAs) The platform reports that on average, members grow gross premiums written through the Risk Exchange by 52% in their first two years. This is not a number the company invented — it is the consequence of what members do when given access to Accelerant's data infrastructure, capacity network, and monitoring tools. No direct MGA testimonials were obtained (the company's website returned 403 errors for deep pages), but the NRR of 135% is itself a form of customer testimony. Members are renewing and expanding, not defecting. The management comment that 15 members have been asked to leave since 2018 — out of 265 active — is exactly the quality discipline one looks for. It indicates the platform is selective, not merely acquisitive.

2. Employee Sentiment The Atlas analysis flagged Glassdoor at 3.3/5 with only 42% positive business outlook. This is below the level I consider comfortable. In my experience, companies that treat employees well build durable organizations; those that do not find their talent leaving at precisely the moment it is most needed. A 3.3 rating is not disqualifying, but it is a yellow flag that I will monitor. For a firm built on proprietary data and specialized underwriting expertise, human capital is not peripheral — it is the moat.

3. Competitors Accelerant operates in a specialist niche within the $109B+ MGA market (per Aon, 2024). Primary competitors are Trisura Group (similar fronting model for MGAs), Ryan Specialty Group (traditional wholesale broker), and technology-enabled insurtechs (At-Bay, Coalition in cyber/professional lines). Accelerant's distinction is the data layer — 57,000 data attributes across its portfolio versus the industry standard of fragmented, opaque data sharing. No competitor is known to have assembled a comparable cross-portfolio data infrastructure. This is a genuine moat, not a marketing claim.

4. Industry/Analyst Views Ten analysts cover ARX with an average "Buy" rating and $19.13 price target — 82% upside from approximately $10.50. The stock debuted +36% at IPO, then fell sharply. This pattern — strong debut, subsequent decline — is common for recently-public companies with PE sponsors who still hold large positions. It does not reflect the business fundamentals. The E&S (excess and surplus) insurance market is expanding as standard carriers retreat from complex, climate-exposed, and cyber risks. Accelerant is positioned squarely in the flow of these displaced risks.

5. Management Reputation CEO Jeff Radke co-founded the company in 2018. The decision to build the data infrastructure first — and only then to add capacity partners — is the mark of a long-range thinker. His disclosure on the earnings call was notably candid: he called out the $39M investment gain explicitly, showed the underlying EBITDA of $66.3M, and set FY26 guidance that implies significant margin compression. A management team seeking to obscure problems would not volunteer this distinction. I view this as a positive indicator on Point 14 of my Fifteen Points.

6. Industry Context The specialty insurance market has structural tailwinds: climate volatility drives E&S demand, cyber risk is growing rapidly, professional liability is increasingly complex, and Lloyd's capacity is rationed. The MGA model — where underwriting expertise is distributed to specialists who know their niches — is gaining share from standard carriers who lack the granular expertise to price these risks. Accelerant sits at the intersection of this structural shift with a platform that reduces information asymmetry between underwriters and capital providers.

7. Hiring Accelerant's careers page returned 403. No direct hiring data obtained. The company has grown from 170 to 265 members (+56%) in 18 months with, presumably, a relatively lean team given the technology-mediated model. This is consistent with platform economics — headcount does not scale linearly with volume.


Fifteen Points Assessment

Point Description Assessment Notes
1 Market potential for products PASS $109B+ MGA market, E&S structural tailwinds, global specialty insurance complexity rising
2 Management determined to develop new products PASS Data attributes 23k→57k; Exchange Services take rate expansion active
3 R&D effectiveness PASS Data infrastructure IS the R&D; 57k attributes = underwriting intelligence flywheel
4 Above-average sales organization PASS NRR 135%; 15-18 net member adds/quarter; +52% member GPW in first 2 years
5 Worthwhile profit margins CONDITIONAL Underlying Q3 EBITDA margin ~24.8%; FY26 guided 5.4% = deliberate mix-shift compression
6 Margin improvement actions PASS Take rate 7.1%→8.0%; Gross Loss Ratio 54.7%→50.1%; structural actions evident
7 Labor relations YELLOW FLAG Glassdoor 3.3/5, 42% positive outlook; not disqualifying but warrants monitoring
8 Executive relations INSUFFICIENT DATA Insufficient public data; founder-CEO with long tenure is a positive proxy
9 Management depth INSUFFICIENT DATA Public company for <1 year; depth of bench not yet visible
10 Cost analysis and controls PASS Improving loss ratios; disciplined member exit record (15 removed since 2018)
11 Industry-specific advantages PASS Investment float ($2.5B portfolio) generates material income; data network effects
12 Long-range profit outlook PASS 3-year strategic target explicit; management making short-term margin sacrifice for platform
13 Future equity financing needs PASS $2.5B investments, $122M debt, clean balance sheet; no dilutive financing needed
14 Management candor under pressure PASS Voluntary disclosure of investment gain component; transparent NRR deceleration discussion
15 Management integrity PASS WITH CAVEAT No integrity red flags; Altamont 79.2% voting control is a governance risk, not an integrity issue — but the structural conflict merits attention

Net assessment: 10 PASS, 2 CONDITIONAL/YELLOW, 2 INSUFFICIENT DATA, 0 FAIL

The Fifteen Points do not disqualify Accelerant. For a company that has been public for less than one year, two "insufficient data" marks are expected and should not penalize an otherwise strong assessment.


Growth Arc Analysis: "Fortunate Because They Are Able"

I have distinguished over many years between two types of growth companies. The first type is "fortunate" — it happens to operate in an expanding industry and rises with the tide. The second type is "fortunate because it is able" — management creates the conditions for growth through superior execution, and the industry tailwind is secondary to that capability.

Accelerant is, in my judgment, the second type.

The specialty insurance market is growing, yes. But many participants in that market are not growing at 74% per year. What Accelerant has built — a proprietary data intelligence layer across 265 specialty insurance books, feeding underwriting decisions for 96 risk capital partners across 22 countries — is not easily replicated. Every quarter that Accelerant remains the hub of this network, the data advantage compounds: more policies → more claims data → better loss prediction → better member selection → better loss ratios → more attractive to capital partners → more capacity to offer members → more members.

The data flywheel is real. 57,000 data attributes (up from 23,000) is not a vanity metric — it represents the accumulated intelligence of six years of specialized underwriting observation, normalized across a diverse portfolio in a way that no individual MGA can replicate.

This is the criterion I weight most heavily. Everything else — the margins, the NRR, the guidance — flows from whether this core flywheel continues to spin.


Management Evaluation

Jeff Radke's handling of Q3 deserves close examination. When a CEO voluntarily strips out a $39M investment gain to present underlying EBITDA of $66.3M — knowing that analysts will use the headline $105M — that is the behavior of someone who wants investors to understand the business, not merely admire the numbers. I have encountered too many managements in my career who do exactly the opposite.

The FY26 guidance is a further test of character. Management is guiding to 5.4% EBITDA margin on $5B+ EWP. This will look terrible in absolute terms — it implies deliberate margin compression as third-party insurer mix rises from 32% to ~42% by year-end and toward 67% over three years. Many analysts will object. The management team is choosing long-range platform economics over short-term margin opticism. This is precisely what I respect.

The risk in this calculation is that third-party insurer economics are inherently lower-margin (commission/fee income ~5% of volume vs. retained underwriting at full spread). The question is whether volume at scale and take-rate expansion produce superior economics in the long run. The math suggests they can: if EWP reaches $5B and Exchange Services take rate holds at ~8.5%, Exchange Services revenue alone = 425Mat 47200M EBITDA — nearly triple today's underlying level. But this requires execution over multiple years.

The Altamont governance question: Altamont Capital controls 79.2% of voting power through 10:1 Class B shares. This is a standard PE governance structure, not an integrity violation. But it is a risk I would be remiss to ignore. Altamont is a private equity firm managing capital for its limited partners. They will exit. The timing and manner of that exit will affect ARX shareholders. The dual-class structure means minority shareholders have limited recourse. This is a structural discount I apply to the position, not a disqualifier.


Quarterly Results — The Numbers

Q3 FY25 (quarter ending September 2025):

Metric Q3 FY24 Q2 FY25 Q3 FY25 YoY QoQ
Revenue [Non-GAAP] $153.7M $219.1M $267.4M +74% +22%
Adj EBITDA [Non-GAAP] $26.1M $63.5M $105.0M*
Underlying Adj EBITDA [Non-GAAP] $66.3M
Underlying EBITDA Margin 17% 29% 24.8%
EWP [Non-GAAP] $795.5M $911.3M $1,042.9M +31% +14.4%
Third-Party DWP [Non-GAAP] ~$227M ~$246M $336M +48% +37%
NRR [Non-GAAP] ~151% 151% 135%
Members 204 248 265 +30% +7%
Exchange Services Take Rate 7.1% 8.0% +90bps
Gross Loss Ratio [Non-GAAP] ~54% 51.8% 50.1% improving improving

*Includes $39M [Non-GAAP] investment gain. Underlying = $66.3M.

Balance sheet health [GAAP]: Total assets $7,859M; Investments $2,554M; Debt $121.9M; Equity $703.8M. Leverage is minimal. The $7.16B in insurance liabilities is matched by investment assets — this is the normal structure of an insurance carrier, not financial distress.

The GAAP net loss of -$1,367M: This is a non-cash accounting artifact from the IPO profits interest distribution. It has zero economic significance. Any analyst who leads with this number has misread the company.


The FY26 Guidance — What It Means

Metric Q3 FY25 Rate FY26 Guidance Implied Change
EWP $1,043M/qtr $5,000M+ FY +28-33% YoY
Third-Party DWP % 32% 42% Mix shift accelerating
Adj EBITDA [Non-GAAP] ~$66M underlying $269M FY +19.6% YoY
Adj EBITDA Margin ~24.8% ~5.4% Deliberate compression

The Q4 guidance (EWP 1, 060−1,100M, up only +3.6% QoQ from +14.4% in Q3) implies a deliberate deceleration as third-party mix ramps. EWP growth is muted near-term because the premium writing is being handed off to third-party insurers — which reduces EWP (Accelerant's retained premium exposure) even as total gross premium on the platform grows faster. This is a nuance that casual readers of the headline EWP number will miss.

The 57−62M Q4 EBITDA guidance (5.5% margin) versus underlying $66.3M in Q3 is the margin compression beginning in real-time.


Conservative Dimensions Assessment

Fisher's four dimensions of conservative investing — applied to ARX:

1. Operational excellence: PASS. Gross loss ratio improving consistently. Member selection discipline evident (15 exits since 2018). Take rate expanding through deliberate product development, not volume-chasing.

2. People quality: CONDITIONAL. CEO Radke — credible founder. Altamont governance creates alignment uncertainty. Glassdoor employee sentiment muted.

3. Business characteristics: PASS. Network effects in data (57k attributes). High switching costs for members (platform is embedded in their underwriting workflow). Recurring fee income (Exchange Services). Float income from investment portfolio.

4. Price vs. fundamentals: PASS at current levels. At 10.50/shareand 3.2B market cap, ARX trades at approximately 2.5x EV/TTM Revenue and ~8x EV/FY26E EBITDA. For a platform business growing revenue at 70%+, this is inexpensive. The stock is at a 66% discount from its IPO-day high. The market is pricing in execution risk on the margin compression story — which is a reasonable concern, but not a fatal one.


Prior Beliefs / Updated Beliefs

Prior beliefs: None (first analysis). Approached with no prior position.

Updated beliefs after this analysis:

  1. The platform economics are genuine. The data flywheel is the moat, not the balance sheet size.
  2. Management quality is above average. Radke's transparency, the member exit discipline, and the voluntary underlying EBITDA disclosure all pass my candor tests.
  3. The FY26 margin compression is a feature, not a bug — if management's mix-shift logic is correct. I am not yet fully persuaded because the execution risk is real.
  4. The Altamont governance structure is a structural overhang that prevents a high-conviction position. PE sponsors do not hold forever.
  5. NRR deceleration (151% → 135%) must be watched closely. Anything below 120% would signal member dissatisfaction.
  6. At current prices, the risk-reward is favorable for a small, initial position. Not a large position. This requires at least 2-3 more quarters of public data.

Verdict

Accelerant Holdings passes the fundamental Fisher tests: it is building a platform with genuine network effects in a structurally growing market, led by a founder-CEO who demonstrates candor under pressure, with a data moat that competitors cannot easily replicate.

The reservations are real. The Altamont governance structure is uncomfortable. The margin compression story requires sustained trust in management's long-range judgment. The NRR deceleration, while still exceptional at 135%, must not continue. The company has too short a public history to warrant deep conviction.

My assessment is: Worthy of an initial, modest position. Not a large position. Watch Q4 FY25 and Q1 FY26 closely.

If the platform is what management says it is — and the preliminary evidence suggests it may be — the time horizon for this investment is many years. The price today, at ~8x FY26E EBITDA on a platform growing at 70%+, leaves adequate room for patience.

If I am wrong about the platform economics, or if Altamont sells in a manner that suppresses the stock, the downside is manageable because of where the current price already sits relative to intrinsic value.

In my experience, the greatest investments are found precisely when a genuinely capable company has been misread by the market — either because the business model is unfamiliar, the near-term numbers look complicated, or a structural overhang depresses sentiment. Accelerant may fit this description. I am not yet fully certain. That is why I recommend a modest initial position, not a full commitment.

Thesis status: INITIATING — Cautiously Constructive | Conviction: 3 of 5


Scuttlebutt Update — Post-Research Addendum

Research completed after initial draft. Material findings that modify the verdict.

The Hadron/Altamont Related-Party Risk — More Serious Than Initially Assessed

The scuttlebutt research surfaced a finding I must address directly. Altamont Capital — which controls 79.2% of voting power in ARX — also owns Hadron Specialty Insurance, which at IPO represented approximately 60% of Accelerant's "third-party" Direct Written Premium. This is not simply a governance concern. It is a fundamental question about the integrity of the business model.

When management reports that "third-party DWP" is growing from 32% to 67% of EWP, and when that metric is the primary strategic indicator of platform evolution, the fact that the largest "third-party" partner is controlled by the controlling shareholder is highly material. The improvement in this metric could reflect genuine platform adoption by independent capital providers — or it could reflect Altamont directing Hadron capacity in ways that support the narrative. Without independence verification, we cannot distinguish between these outcomes.

In my experience, the test is not whether a related-party transaction is conducted at market terms — management always believes it is. The test is whether the minority shareholder has the means to verify and enforce arm's-length behavior. Here, with 79.2% voting control held by the party in question, that enforcement mechanism does not exist.

This is the concern that led one independent analyst (Messerli Capital) to downgrade ARX to the "too hard" pile in January 2026, explicitly citing "execution risk and governance concerns." I concur with the substance of that caution, though I would frame it more precisely: the risk is not that management is acting with bad faith, but that the structure makes verification impossible and correction mechanisms unavailable.

Revised scuttlebutt findings summary:

Category Finding Weight
MGA customer sentiment No independent data; curated metrics positive (39% avg GWP growth); B2B opacity Neutral
Employee sentiment 3.3/5 Glassdoor, 42% positive outlook, nepotism explicitly cited Negative
Competitive position Novel model, no true comp; Trisura/Ryan Specialty as partial competitors Positive
Analyst consensus 8 analysts, avg Buy, avg PT ~$21; Messerli base case = 6% IRR only Mixed
Management Radke credible operator, 30+ years; governance structure problematic Bifurcated
Industry backdrop E&S secular tailwind strong; AM Best signaling cycle maturation, rate softening Favorable but moderating
Hiring AI/ML product managers, European MGA recruitment, data quality investment Positive growth signal
Hadron related-party Altamont owns fronting insurer representing ~60% of "third-party" DWP at IPO Materially negative

Revised Verdict

The Hadron finding reduces my conviction from 3 to 2.5 of 5. The platform economics remain genuine in the abstract. The data flywheel logic is sound. Radke's operational credibility holds. But the structural architecture — PE sponsor controlling both the platform and its largest "third-party" capital provider, with no governance recourse for minority shareholders — introduces a verification problem that I cannot dismiss.

This does not disqualify the investment. It does mean the position must remain small until: (a) Altamont's Hadron relationship is disclosed more transparently in future filings, (b) genuinely independent third-party insurer share (excluding Hadron) shows growth, and (c) Altamont's exit timeline becomes clearer.

The price — ~2.5x EV/TTM Revenue — still provides margin of safety for a modest position. But "modest" now means 1-2%, not 3-4%.

Final Conviction: 2.5 of 5 | Thesis: INITIATING — Cautious, Small Position Only