Thesis

The Long Tail Has No Price

On isolated views, manufactured markets, and liquidity as an underwriting business

I. Markets bundle. Beliefs don't.

The most common trade in finance is a compromise. A trader does the research and arrives at a precise belief about one thing — a ruling, a regulatory decision, a data release, a contractual milestone — and then goes looking for an instrument. What they find prices something adjacent: the company, the sector, the index, the volatility surface. The belief is about one number. The instrument is about ten. To express the one, the trader must warehouse the other nine.

This is not a niche complaint. A fund believes a merger closes but that the market's implied timeline is wrong. A credit desk holds a note whose value hinges on a milestone that has no market of its own, so the desk marks a bundle and hopes. An energy trader has a view on a single permitting decision, not on the entire forward curve it sits inside. A macro pod believes one print will surprise and has no view at all on the policy reaction that follows. In every case the underlying event is well-defined, resolvable, and professionally researched — and in every case the nearest instrument bundles it with beta, rates, idiosyncratic noise, and every other fact that happens to resolve inside the same expiry window. The premium paid or collected is mostly rent on exposures nobody wanted. Both sides of the trade pay it: one side warehouses tails it had no view on, and its counterparty warehouses the inverse.

Prediction markets were supposed to be the answer, and in one sense they delivered: they proved the demand for direct event exposure is enormous. But the volume concentrated where the audience is — elections, sports, a handful of macro headlines — and the long tail of events with real economic consequence and narrow audiences remains unpriced. The standard explanation is that these markets are "too illiquid to exist."

That rationale is only partially sound, and the half that's not is quite interesting.

II. The misdiagnosis

It is true that there is no natural, continuous, two-sided market in a bespoke event. A binary has no organic other side; between events there is nothing to trade, and at events the demand is episodic, concentrated, and research-driven. Anyone waiting for a crowd to show up and stream quotes into a long-tail event market will wait forever, and the venues that tried have the empty order books to show for it.

The error is in the conclusion. What cannot exist is a long-tail event market that looks like the markets we already have. What can exist is a manufactured one — a market in which the quote is produced by a mechanism rather than a counterparty, and size is provided by a professional warehouse that is paid to hold risk it cannot hedge.

The distinction that makes this tractable is that price discovery and liquidity are different problems with different economics, and conflating them is the canonical failure of event-market venues. Price discovery — a live, credible number that anyone can trade against — can be manufactured by an automated market maker with a bounded subsidy. Liquidity in the sense a trading desk means it — someone willing to absorb size at a usable spread — cannot be manufactured; it has to be underwritten. The natural counterparty to a long-tail event is not a market maker clipping spreads between hedged legs, because across an event there is no hedge to clip against. It is an underwriter: capitalized to warehouse the risk through resolution, paid a premium for doing so, and made viable not by any single market but by a diversified book of them, where per-event risk falls toward a correlation floor as the book widens. These are insurance economics, not market-making economics. Reinsurers do not hedge hurricanes. They diversify them, and they price the premium.

Once liquidity is correctly defined as underwriting, the "too illiquid" objection dissolves into two honest, answerable questions: what premium does the warehouse require, and does real hedging and expressive demand show up at a spread that pays it? Those are measurable facts about a market, not verdicts about an asset class.

III. Markets without anchors

Long-tail events pose a second problem, quieter and more fundamental than liquidity. Every market that has ever priced anything borrows its discipline from an external anchor that eventually, forcibly arrives. A listed equity is disciplined continuously — the next print is seconds away. A weekend perpetual prices self-referentially until Monday's reopen grades it. A pre-IPO perpetual drifts for months, but a listing print eventually arrives and settles the argument. In each case the market can wander, but something outside it returns.

An event market is the degenerate case: nothing ever returns. Its only contact with external truth is the resolution itself, and resolution does not re-anchor the price — it terminates the market. Along the entire life of the contract, no outside number arrives to grade the inside one.

So the discipline cannot be borrowed. It has to be built into the pricing mechanism itself, and it is worth stating exactly what that requires. A quote must always exist, expressible against the mechanism with no matched counterparty, or the market fails at inception. The sponsor's cost of providing that quote must be bounded and provable, or the quote's budget is not credible. Money alone must be unable to set the price — a subsidy should buy depth, never a level — so that the resting price moves only on information. And settlement must be an identity rather than a convention, so that everything built on top of the price resolves to the same fact. When those properties hold, the number is credible not because a crowd polices it, but because manipulation is structurally unprofitable and honesty is structurally free. That is what makes this infrastructure rather than a product: the mechanism, not the operator, is the thing being trusted.

IV. Leverage must be underwritten

The other half of what traders actually want is capital efficiency, and here the on-chain decade has already run the experiment. Perpetual futures became the most successful derivative innovation of their generation because they deliver isolated, leveraged, continuous exposure tethered to a reference price by funding. The demand for expressing a view without posting the full notional is not speculative excess; it is how professional capital allocates attention across many positions at once.

But event exposure resists the standard construction, and the reason is an identity, not an engineering gap. In a zero-sum book on a binary outcome, symmetric leverage cancels at settlement: both sides scale their margin down together, the pot scales with them, and the winner's return multiple collapses back to the unlevered one. Settlement leverage on an event cannot be manufactured inside the book — it can only be underwritten from outside it, by a party willing to over-collateralize relative to its own risk and be paid for it. The funding rate is that payment. Which means the venue's economics converge from two independent directions onto the same counterparty: the liquidity problem demands a paid warehouse, and the leverage problem demands one too. A long-tail event venue is, at bottom, an underwriter of unhedgeable dated risk — and everything else is the machinery that keeps its price honest and its losses bounded.

The same logic disciplines the instruments themselves. Across an event there is no continuous path to hedge through; the standard volatility toolkit, built for diffusion, does not survive contact with a price whose variance arrives in a single jump. Instruments for events have to be designed for the jump — contracts on the magnitude of the revision — rather than borrowed from markets that move smoothly. Taking that seriously is the difference between building the right instruments and shipping a familiar-looking one that a sophisticated counterparty would immediately recognize as degenerate.

V. The price is the product

What does the world get when this works? Not primarily a place to bet. It gets a number that did not exist before: a continuously updated, manipulation-resistant probability for an event that previously traded only as a rumor inside a bundle. That number is a reusable primitive. It marks the contingent paper that desks currently mark by hope. It feeds risk systems that currently carry event exposure as an unmodeled residual. It gives decision-makers — allocators, operators, policymakers — a live, credible read on questions that today are answered by whichever expert spoke last.

This was always the serious case for prediction markets: not the volume, the information. The critique of the existing venues is not that speculation is unseemly; it is that the sportsbook model concentrates all of its machinery on the events that least need pricing, while the long tail — where an honest number would actually change decisions — goes unserved because it cannot supply a crowd. A venue built on manufactured discovery and underwritten liquidity does not need a crowd. It needs a mechanism worth trusting, a warehouse worth paying, and a stream of participants with genuine views. That is a different, smaller, more demanding requirement — and it is the one the long tail can actually meet.

VI. Where this starts

A thesis this general still has to land somewhere specific, and the discipline is to start where events are cleanest: scheduled, hard-dated, binary, publicly resolvable, and researched by professionals with real conviction. Our first vertical is biotech catalyst risk — regulatory decisions and clinical readouts are among the cleanest belief-revision events in public markets, with institutional demand already routed through instruments that bundle them badly. But the vertical is an instance, not the thesis. The venue takes no view and originates no risk; it aggregates the belief revisions that already exist in specialist hands and lets them be expressed undiluted. Trade the event, not the bundle.

What remains genuinely open, we state plainly, because the credibility of everything above depends on it: mechanisms can be proven, but markets have to be measured. Whether informed participation secures the price, and whether real demand arrives at a spread that pays the warehouse, are empirical facts about participants — not properties of the design. The infrastructure exists to answer those questions honestly, one measured market at a time. The long tail has no price today. The claim of this document is not that pricing it is easy. It is that, for the first time, it is an engineering and underwriting problem rather than a wish.