Two-Thirds of the Optimal Transactions Happen When the Market Doesn’t Trust the Information. Insurance Runs on That Missing Third.

Jeff Radke got into insurance the way a lot of second-generation people do: he came home covered head to toe in dirt from a summer job laying sod, and found Clem Dwyer — an old friend of his father’s from Guy Carpenter, the reinsurance broker — sitting in his father’s kitchen. Dwyer offered him a summer job at Guy Carpenter for $7.50 an hour. Air conditioning and no more dirt won easily. His father, Jerry Radke, was already in reinsurance, making Jeff second-generation in the business.

Accelerant’s founding thesis came from watching a broader industry trend: genuinely talented underwriters were leaving large, monolithic insurance companies to start their own MGAs (called program administrators in the US) — a pattern Jeff considers part of a wider economic shift where technology increasingly lets skilled individuals operate independently. What Accelerant observed was that the insurance companies theoretically positioned to serve those new MGAs were, in Jeff’s blunt words, doing “a terrible job” — not meeting their needs, sometimes actively creating new problems instead of solving old ones.

In Episode 82 of InsurTechTalk, Jeff and I covered why Accelerant deliberately shares all of its data with both sides of its platform rather than guarding an “information edge,” how a specialist in poultry house insurance became one of the company’s proudest investments, and why cyber remains genuinely hard to underwrite confidently.

About Jeff Radke

Jeff Radke is CEO and co-founder of Accelerant, a platform matching specialty underwriters (MGAs and program administrators) with risk capital providers (reinsurers and capital markets investors). Accelerant operates across 16 countries with roughly 178 different products, all structured as diversified, non-catastrophe-heavy small-to-medium commercial business. The company has no internal departments or silos — its roughly 200 employees are organized entirely around serving both sides of the platform: the underwriting “members” and the risk capital providers.

What Accelerant Actually Does

Jeff’s own simplified framing: Accelerant matches specialty underwriters with the right kind of risk capital, and gets paid by the risk capital providers to produce and manage that portfolio. The underlying insight driving the model: individual MGA portfolios, taken alone, are often too small or too narrow for a reinsurer to want direct exposure to — but aggregated together, a genuinely diversified book of small-to-medium commercial business (explicitly not catastrophe-heavy, not satellite risk, not refineries) becomes a portfolio with attractive volatility and profitability characteristics that a reinsurer’s own book actually benefits from holding.

Information Asymmetry as the Actual Root Problem

This was the intellectual core of the conversation, and it’s a genuinely sharp reframe of what’s usually wrong in insurance distribution.

Jeff was blunt about a pattern he hears constantly in this space: nearly every insurtech founder claims to have “an information advantage” or “better data than others.” His pointed challenge to that framing: Accelerant does the opposite. Rather than guarding a proprietary data edge, Accelerant shares every piece of information it has with both sides of the platform simultaneously — the MGA/underwriter and the risk capital provider look at the exact same analytical platform, with nothing withheld from either side.

His theoretical grounding for why this matters: he referenced the 2020 Nobel Prize in Economics, awarded specifically for research on how markets break down under severe information asymmetry — with a striking finding he paraphrased: markets without confidence in fair information flow achieve only roughly two-thirds of their optimal transaction volume. Jeff’s read: that finding describes insurance markets almost perfectly. His poker analogy: if you sit down at a table and don’t know who the sucker is after two hands, it’s you — and he spent much of his earlier career genuinely worried about being on the wrong side of exactly that asymmetry in insurance transactions.

Accelerant’s answer, which the company calls “aggressive transparency”: actively pushing shared data into both the MGA’s and the capital provider’s consciousness, rather than passively making it available. Jeff’s framing of the second half of the “special sauce”: Accelerant has no internal silos or departments — no separate “programs department” or “MGA division” — meaning there’s structurally less internal friction preventing information from flowing cleanly to where it’s needed.

How Fair Return Actually Gets Split

Jeff described Accelerant’s economic philosophy precisely: each product carries an internally-defined fair return target for the capital supporting it, calibrated to that product’s actual volatility and correlation profile. Where a product’s portfolio performs better than that fair return, the excess value flows to the member (the MGA or underwriter) that produced it — not retained by Accelerant. His framing to risk capital providers is correspondingly direct: here’s the portfolio, here’s what we believe is a fair return on it, and we’re prepared to walk through all 385,000 outstanding US policies individually if that level of verification is wanted.

Running Experiments Where Traditional Insurers Say No

Jeff’s example here was the clearest illustration of Accelerant’s actual operating philosophy in practice. A relatively new (not brand-new startup) member company came to Accelerant with strong embedded-insurance technology and a genuine skill for understanding what a distribution platform’s owner actually needed — but the coverage category itself was brand new, with no historical loss data to price against.

His description of the standard insurance industry reflex in that situation: “there’s no history, we can’t help you” — which he considers structurally wrong, not just unhelpful. Accelerant’s alternative: run the experiment. Get pricing directionally right, expect some early losses, and treat those early losses as effectively a marketing or product development cost rather than a failure — then correct pricing iteratively back toward the fair-return target once real data accumulates.

His explicit point on why this is organizationally hard to do elsewhere: running that kind of experiment requires an organization genuinely joined up around shared outcomes, not fragmented across separate P&Ls where a failed early bet triggers internal blame (“the sharks come out,” in his phrase). He also credited Chief Underwriting Officer Frank O’Neill’s stated default posture — trying to find a way to say yes — as a cultural discipline that changes outcomes on its own, independent of any specific pricing methodology.

Flywheel Re: Opening Capacity to Non-Traditional Capital

Jeff described Flywheel Re as Accelerant’s answer to a specific capital-sourcing constraint: a meaningful share of the risk capital Accelerant wanted access to came from capital markets investors who had no licensed reinsurer status and often no direct insurance underwriting experience at all — but who found the category genuinely attractive as a portfolio diversifier, since it’s uncorrelated with interest rates or equity markets. Flywheel Re was built specifically to let that class of capital participate in Accelerant’s portfolio economics despite lacking a traditional reinsurance license.

What members actually get from this structure, per Jeff: stable, long-term capacity at predictable pricing — arguably the single thing MGAs and program administrators value most, since unpredictable capacity availability is one of the most disruptive risks to running a specialty underwriting business.

The Poultry House Story

This was Jeff’s favorite concrete example of Accelerant’s member model working exactly as intended, and he was visibly enthusiastic describing it. A specialist named Will — who Jeff says knows more about poultry house insurance than perhaps anyone in the world — runs a small team of roughly 10-15 colleagues genuinely passionate about the same extremely narrow niche.

Jeff’s framing of why this matters: conventional wisdom in the industry holds that poultry business is a terrible risk category — bad losses across the board — except, as Accelerant’s Head of Portfolio Hugh Burgess puts it, for this specific poultry business, because Will’s team has produced consistently strong results over multiple years through genuine specialized expertise, loss control discipline, and deep category knowledge most generalist underwriters simply don’t have.

Accelerant didn’t just provide capacity — it invested directly in Will’s company to help him expand faster than organic growth alone would allow. Jeff’s closing reflection: watching Will’s monthly numbers come in and seeing a business that had genuine potential actually flourish is, in his words, one of the most rewarding parts of the job — a genuine, specific illustration of what “supporting talented underwriting teams” means in practice rather than as an abstract mission statement.

The Real Investment Thesis: Finding Talent, Then Letting Them Own It

Jeff’s framing of what Accelerant is actually looking for when it evaluates a new member, stripped to its essence: exceptional management and underwriting teams, full stop — genre-agnostic, whether that expertise is in Spanish sports club insurance or poultry houses in the southeastern US. His extension of that thesis: the deeper differentiator isn’t just finding talent, it’s structuring genuine ownership for that talent — everyone at Accelerant itself is an owner, and Jeff considers that ownership mentality, not just skill, the reason members like Will’s team perform the way they do. His pointed contrast: an “owner mentality” behaves fundamentally differently than an employee mentality inside a traditional corporate division.

Cyber Insurance: Genuinely Hard, and Honest About Why

Asked for his view on the ongoing cyber capacity crunch, Jeff was direct that he doesn’t consider it solved, and explained the two specific structural reasons Accelerant finds it genuinely difficult to underwrite confidently, even while acknowledging other teams doing serious, credible work in the space:

• Coverage uncertainty at the point of claim. Nobody fully knows how courts will ultimately interpret cyber coverage language written today, against legal and regulatory frameworks that will exist years from now when a claim is actually litigated — a genuine forward-looking interpretation risk baked into every policy written today
• Aggregation risk that behaves unpredictably. Traditional property risk aggregation is at least conceptually clear — insuring two buildings in the same region means understanding a shared storm exposure. Cyber risk aggregation is murkier: Jeff’s concern is that individual cyber exposures may not remain statistically independent the way traditional insurance theory assumes exposure units should be — and once exposure units become dependent rather than independent, the fundamental math underlying insurance pricing starts to genuinely strain, not just become harder to calibrate

Advice: Measure What Matters, Not What’s Easy to Measure

Asked for a closing recommendation, Jeff pointed to Measure What Matters (visible on his own bookshelf during the call) — the book on OKRs (objectives and key results) that Accelerant uses extensively internally. His framing of why it matters specifically for a company like Accelerant: it’s easy to get lost in KPI tracking that measures internal activity nobody outside the organization actually cares about. His sailing analogy: it’s easy to keep your head inside the boat, focused on internal minutiae, when the metrics that actually matter concern the people outside your organization — members and risk capital partners — not just internal operational tidiness.

Key Takeaways

• Information asymmetry, not lack of technology or data sophistication, is the root cause Jeff identifies behind insurance’s structurally poor customer experience — grounded directly in Nobel Prize-winning economic research on market breakdown under information opacity
• Accelerant’s “aggressive transparency” — sharing identical data with both underwriting members and risk capital providers — is a deliberate inversion of the “we have an information edge” positioning common across insurtech
• Running genuine pricing experiments on brand-new coverage categories, treating early losses as product development cost rather than failure, requires an organization without internal silos or separate P&Ls that punish early missteps
• Flywheel Re demonstrates a structural solution to a real capital access problem: letting non-traditional, unlicensed capital markets investors participate in specialty insurance portfolio economics they’d otherwise be excluded from
• The poultry house example illustrates Accelerant’s actual thesis in practice — genuine category expertise can outperform “conventional wisdom” about a supposedly bad line of business, and is worth direct investment, not just capacity support
• Cyber insurance remains genuinely difficult to underwrite confidently for two distinct structural reasons: forward-looking legal interpretation uncertainty, and the open question of whether cyber exposure units are truly independent or secretly correlated in ways traditional insurance math doesn’t yet account for
• Structuring genuine ownership for talented underwriting teams, not just providing capital or capacity, is Jeff’s stated differentiator for why Accelerant’s members outperform — an owner mentality behaves differently than an employee mentality inside a traditional corporate silo