Naive to Native: The Junior Tranche Problem

The Word We Use for Early Money

There is a word the market keeps for people who arrive too early. It calls them naive.

The word is efficient. It sorts. The founder who remortgages the house, the angel who writes the first check into a company with no revenue and no comparables, the backer who funds the thing everyone else can plainly see is too soon: all of them collect the same label, and the label does something quietly useful for the people applying it. It lets them decline without underwriting. Naive is a verdict that saves you the trouble of forming one.

Sometimes the verdict is correct. Some early money really is foolish. There is no honest version of this essay that pretends every premature bet is secretly wise, and the people selling that story are running a different hustle from mine.

But notice what the market cannot do. It cannot tell the careful, high-conviction, well-reasoned early bet apart from the genuine fool. It has one bucket for both. The disciplined backer who has priced the risk and decided to bear it, and the gambler who has priced nothing, get sorted into the same pile and quoted the same terms, which is to say no terms at all. The market does not distinguish kinds of early risk. It only knows early and late.

So who is actually being naive here?

If you have a single category for a whole spectrum of behavior, and you treat everything in that category identically, you are not being prudent. You are being coarse. The naivety is not in the person who shows up early with conviction. It is in a market that lacks the resolution to tell conviction from foolishness, and so prices both as foolishness. The early bettor is not the naive party in this story. The instrument is.

This essay is about fixing the instrument.

What Native Means

To be native to capital markets is to have a recognized position in the stack. A slot the rules acknowledge. A claim with its own price, its own eligibility, its own line in the documents, its own buyers who know exactly what they are buying. Native does not mean welcomed. It does not mean subsidized. It means legible and legitimate: the system has a place for you, and that place has a name and a number.

The opposite of native is not unwelcome. It is unhoused. The unhoused risk is the one with no slot, no clean price, no defined buyer, the risk that can only be funded by someone willing to take the entire bundle or nothing at all. Most early capital is unhoused in exactly this sense. It is not that nobody wants to bear the risk. It is that the structure offers no seat shaped to bear it from.

One thing I want to establish: native does not mean safe. A native seat can be the most dangerous seat in the building. The first-loss position is native. It has a name, a price, and a buyer. It is also the seat that burns first. When I say we should give early risk a native home, I am not saying we should make it safe. I am saying we should make it legible, priceable, and legitimate, so that the people who can actually bear it can find it, price it, and choose it on purpose.

Keep that distinction close. Almost everyone who gets this wrong gets it wrong by quietly sliding from native to safe.

Capital Is Structurally Late

The standard complaint about capital is that it is cowardly. The brave money has dried up, the gatekeepers have no nerve, somebody should be bolder. I used to believe a version of this, and it is wrong, and its wrongness is the most important thing to understand about why the problem persists.

Capital is not late because the people allocating it are timid. Capital is late because the mandates that govern it are built to be late. A pension fund, an endowment, a regulated insurer: each of these runs on a mandate that exists precisely to keep it out of the unproven-beginning bet, the bet where all of the risk sits on the innovator and none of it has been priced by anyone yet. That mandate is not a personality flaw. It is a design choice, made on purpose, often by law, to protect the people whose money it actually is.

This matters enormously, because it tells you the cure. If lateness were a failure of courage, the fix would be exhortation. Be braver. Take more risk. And exhortation has never worked, not once, because you cannot shame a mandate into changing its shape. But if lateness is a failure of design, then it has a design fix. You do not need to make the timid brave. You need to build a vehicle the cautious are allowed to ride.

Here is the reframe the rest of this essay turns on. The problem is not that there is too little appetite for risk in the world. There is enormous appetite, sitting in a thousand different shapes, each one constrained to its own narrow band. The problem is that the appetite cannot find the risk that matches it, because the vehicles we offer are too coarse to make the match. The risk-loving capital and the risk that needs it are in the same room, and the furniture is built so they can never sit together.

Risk Has a Shape

To see the coarseness, you have to see that risk is not a single quantity. It has a shape.

Picture an axis. At one end, principal-protected: you will get your money back, your upside is modest, your floor is solid. At the other end, first-loss equity: you are the first to be wiped out and the last to be paid, and in exchange you own all of the upside above the line. In between sits everything else, the whole graded territory of mezzanine, of senior-but-not-safest, of subordinated-but-not-equity.

Now the constraint. A fund, a bond, an ordinary equity stake: each of these forces you to take the entire shape as one bundle. You cannot buy the safe eighty percent of a credit fund and leave the dangerous tail to someone else. You cannot buy only the upside above net asset value and let another party hold the principal. The vehicle hands you the whole risk profile, top to bottom, and asks you to swallow it whole or walk away.

And most capital, by mandate, can only live at one point on that axis. The insurer can hold the senior band and nothing below it. The specialist credit fund can hold the mezzanine. The risk-seeking pool can hold the bottom. Each is permitted its own slice and forbidden the rest. So when an asset arrives whose natural risk does not match the single band that some available pool of capital is allowed to occupy, the asset does not get a worse price. It gets no price. It simply does not get funded, not because no capital wanted its risk, but because no vehicle existed to hand that capital its slice and only its slice.

That is the gap. Everything that follows is an attempt to build furniture fine enough to close it.

The Primitives

Here is where I get specific, because this is the part that is actually mine to write, and the part where the hand-waving usually begins.

The first primitive is the one this essay is named for. Tranching, and the loss waterfall beneath it. You take an asset, or a pool of assets, and you carve the claims on it into layers. The senior layer is paid first and absorbs loss last. The junior layer is paid last and absorbs loss first. The senior buys its safety not from thin air but from the junior, which stands beneath it and takes the first hit so the senior never has to. The insight that makes tranching profound, and not merely clever, is this: it lets risk-averse capital touch an asset it could never have held whole. The insurer who could never own the raw pool can own the senior tranche of it, because someone else has agreed to stand in front.

The second primitive splits time from upside. Principal and yield separation, in the jargon. You take a cashflow-bearing asset and you cut it into two claims: one that says I want my money back at maturity, and one that says I want the income and the appreciation along the way. These were always two different desires, bundled together by accident of the instrument. Separate them and you have two assets, each with its own natural buyer, where before you had one asset that half-satisfied two people.

The third primitive is the isolated, parameterized market. Instead of pooling every risk into one blended soup, you define a single risk band precisely: this collateral, this maturity, this rate, this liquidation threshold, and nothing else mixed in. The newest on-chain credit infrastructure does exactly this, building intent-based, fixed-rate, fixed-term markets where the terms are quoted by the participants and the risk of each market is isolated from every other. It is the opposite of the giant commingled pool. It is a thousand precise small rooms instead of one large undifferentiated hall.

The fourth primitive is the one that turns the other three from a finance-textbook exercise into something genuinely new, and it is the reason I do this on-chain at all. Programmable eligibility. Each band, each tranche, each claim can carry its own rules about who is permitted to hold it, and those rules are enforced at the level of the token itself, not in a side agreement that someone has to remember to police. The safe senior tranche can be made holdable only by capital that is permitted to hold safe senior things. The dangerous junior can be made holdable only by qualified risk-bearers who are allowed that exposure. The eligibility travels with the asset and enforces itself. This is the thing traditional structuring has never been able to do cleanly, because in the old world eligibility lives in paper and trust, and paper and trust do not scale and do not enforce themselves at three in the morning.

Four primitives. Tranching, principal and yield separation, isolated parameterized markets, programmable eligibility. Together they are a set of chisels fine enough to carve risk into the shapes that waiting capital is actually allowed to hold.

What Changes On-Chain

I should say plainly what the chain adds, and only what it actually adds, because the genre is thick with people claiming it adds magic.

It adds granularity. The marginal cost of cutting one more slice falls close to zero, so you can slice finely enough to match bands that were never worth the legal expense of matching before.

It adds composability. A senior tranche of one pool can become collateral inside another, claims can be recombined into new claims, the pieces snap together in ways that paper instruments frozen in fund wrappers never could. This is real, and it carries real danger, which I will get to, because composability is also how a small fire becomes a system.

It adds transparency of the waterfall. The rules of loss allocation are written in code and visible in real time, rather than buried in a prospectus that updates quarterly if you are lucky. You can watch the water fall.

It adds native secondary liquidity. Each band can have its own market instead of being locked inside a vehicle until maturity, so the slice you bought is a slice you can sell.

And it adds the programmable eligibility from the section above, which is the one I would keep if I could keep only one.

That is the honest list. It is a real list. Now here is the warning label it comes with, because the next section is where I stop selling and start confessing.

Native Is Not Safe

Tranching is the technology that produced 2008. I am not going to pretend otherwise, and any account of this subject that does not put that sentence near the center is not to be trusted.

So let me separate, as cleanly as I can, what is genuinely different now from what is not, because the difference is narrower than the enthusiasts claim and wider than the cynics admit.

What is genuinely different is legibility. A great deal of the 2008 catastrophe was opacity. Nobody could see what was inside the instruments, the waterfalls were buried, the same collateral was pledged in places nobody could trace, and the ratings were a fiction layered on a fiction. On-chain structuring attacks exactly that failure. The waterfall is visible. The collateral is traceable. The rehypothecation, if it happens, happens in the open. That part of the 2008 disease, the opacity, is real, and on-chain transparency is a real treatment for it.

And here is what is not different, what I need you to concede with me, because the whole credibility of this project depends on conceding it. Tranching redistributes risk. It does not destroy it. It moves the risk to whoever holds the junior; it does not make the risk smaller. And the thing that actually kills tranched structures is not opacity. It is correlation. The senior tranches of 2008 detonated because everyone assumed the underlying assets were independent of one another, and they were not, and when the correlation revealed itself the cushion that was supposed to protect the senior simply was not thick enough, because the loss did not arrive one borrower at a time the way the model assumed. It arrived all at once.

On-chain transparency does nothing about correlation. Nothing. If your collateral pool is, to take an example uncomfortably close to my own work, entirely Kenyan fixed income, then a single sovereign shock takes every tranche in the structure at the same moment, senior and junior together, and your beautiful real-time on-chain waterfall does not save anyone. It just renders the fire in higher resolution. You will be able to watch, with perfect clarity and admirable latency, every tranche burn at once.

So when I say native, hold me to the definition. Native means placed, priced, legitimate. It does not mean safe. The technology improves legibility and access. It does not amend the laws of risk, and anyone who tells you it does is selling you the 2008 lie in a nicer font.

The Junior Tranche Problem

Now we arrive at the actual problem, the one I named this essay for, and it is not a technical problem. The technical problems are all solved. This one is harder than technical.

Tranching only works if someone holds the junior. The entire architecture, the senior safety, the matched bands, the capital finally finding its shape, all of it rests on the existence of a party willing to stand at the bottom and absorb the first loss. The structure can make that role beautifully legible. It can price it, gate it, make it tradeable, render it in real time. What the structure cannot do is conjure the risk-bearer into existence. Legibility is not creation. You can build the most exquisite seat in the world and it does nothing at all until someone agrees to sit in it.

So let me reframe what the junior tranche actually is, because the usual framing is the source of all the trouble. The junior tranche is not a high-yield product. It is a paid job.

The yield on the junior is not a gift, and it is not free money. It is a salary. It is the compensation for doing a specific and difficult piece of work: absorbing the first loss, and pricing the underlying correctly enough that the compensation turns out to be adequate to the risk. The yield has to be high enough to recruit someone competent to do that job, and no higher, because if it is higher than the risk requires you are either overpaying and quietly bleeding the structure, or, more often, you are signaling that the structure is desperate and cannot find an honest taker at an honest price.

And not everyone can do the job. The junior seat requires three things that cannot be faked. It requires a balance sheet that can absorb the first loss without that loss becoming fatal, because the entire point of the junior is to take the hit, and a junior holder who is destroyed by the hit was never the right holder. It requires the genuine ability to price the underlying, because the junior is where mispricing is lethal: the senior holder can be a little lazy about the collateral and survive on the cushion beneath, but the junior holder who misprices the pool is the one who personally meets the correlation everyone else ignored. And it requires a mandate that actually permits first-loss exposure, because plenty of capital is forbidden this seat regardless of how attractive the yield, and forbidden for good reason.

The cleanest holder of the junior, the one that makes the whole structure honest, is the originator itself. When the party who assembled the pool keeps the first-loss piece, the junior stops being merely a yield position and becomes a bond posted on their own underwriting. It is skin in the game made structural. It says, in a language the market can read, I believe my own work enough to be the first to lose if I am wrong. That is what makes the senior credible to everyone standing above it. The junior held by the originator is not just a tranche. It is a signature.

The Retail Inversion

Here is the tempting wrong answer, and I want to confront it directly, because it is the answer my own critics will reach for and the answer that, dressed up nicely, gets repeated in rooms where it should be laughed out.

The instinct goes like this. We have a junior tranche that needs a holder. We have an enormous pool of underused retail capital sitting on the sidelines, locked out of these returns. Why not match them. Let the crowd hold the first-loss piece, democratize the high yield, open the junior to everyone.

This is precisely the inversion the entire structure exists to prevent.

The whole logic of senior and junior is that the sophisticated, well-capitalized, loss-absorbing party stands in front so that everyone behind them is protected. To put retail in the junior is to take the person least able to price the risk and least able to survive the loss, and to seat them in exactly the chair that prices and absorbs first. That is not democratization. That is the subprime mis-selling of 2008 rebuilt with better graphics. The crowd did not need the chain to be sold first-loss risk it could not price; it managed that catastrophe perfectly well on paper. We should not be proud of automating it.

But there is a real idea buried in the instinct, and it is the exact mirror image of the wrong one. Retail is indeed an underused market. It is just underused at the other end of the axis. The thing retail has been locked out of is not the junior. It is the senior: the safe, credit-backed, predictable yield that sits on top of the waterfall, insulated by the junior beneath it. Retail has always been shut out of the senior claim, not because retail should not have it, but because the vehicles to deliver it safely at scale, with eligibility actually enforced, did not exist. Programmable eligibility is what changes that. The junior stays with the qualified risk-bearer who can do the job. The senior, made safe by that junior, becomes something the gate can finally open to a vast retail base that the old world kept outside.

And now let me bring this home, to the ground I actually stand on, because this is where the argument stops being general and becomes mine. In the markets I work in, retail already bears junior-shaped risk. It bears it constantly, blindly, and without a shred of the legibility or protection a real structure would give it. The informal credit sector across the continent, the SACCOs, the chamas, the table-banking circles, the mobile lenders charging rates that would be called obscene if they were ever made legible, all of it is retail capital sitting in a de facto first-loss position every single day. The risk is already being borne. It is simply being borne in the dark, unpriced, unprotected, and uncompensated at anything resembling a fair rate.

So the opportunity is not, and never was, to push retail into formal junior tranches. The informal sector is already the junior tranche, badly. The opportunity is twofold, and it points the other way. Give retail clean, gated access to the senior yield it was always denied. And make the junior-shaped risk that retail is already carrying, in the chamas and the circles and the mobile loan books, finally legible and fairly priced, so that the people bearing it can see what they hold and be paid honestly for holding it. The unlock is not new first-loss bodies. It is light on the first loss that is already being borne.

Two Answers to One Question

There is a question underneath this entire essay, and naming it cleanly is worth more than any amount of cleverness. The question is: who absorbs the first loss.

There are two real answers, and they are genuinely different, and most of the confusion in this field comes from blurring them together.

The first answer is tranching. You recruit a third party, the junior holder, to stand at the bottom and absorb the first loss in exchange for the salary we just discussed. This is the securitization model. It works whenever you can find, or be, a competent junior. It fails whenever you cannot.

The second answer is overcollateralization, and it is the model most on-chain lending actually runs on, though almost nobody describes it in these terms. In an overcollateralized lending market, there is no third-party junior holder at all. Instead, the borrower posts their own first-loss buffer up front, in the form of collateral worth more than the loan, and if that collateral falls toward the value of the debt it is liquidated before the lender is ever touched. The first-loss bearer is not a junior investor. It is the borrower, insuring their own loan with their own capital. This is the quiet genius of the modern fixed-rate, fixed-term on-chain credit markets: they sidestep the junior tranche problem entirely, not by solving the recruitment of a first-loss holder, but by making the borrower self-insure so that no separate first-loss holder is needed.

The two models trade against each other cleanly. Tranching is capital-efficient and works on assets you cannot easily collateralize, but it requires you to find a junior. Overcollateralization needs no junior, but it is capital-inefficient and works only on assets you can post as collateral. Neither one is the answer. They are two answers, for two different situations, to the same question.

And this is the practical thing every builder in this space has to internalize, because I have watched people assume otherwise and design themselves straight into a corner. If you build on the overcollateralized fixed-term primitives, you get the safe, gated, senior-style leg natively. You do not get a junior tranche thrown in. The waterfall, the senior-over-junior subordination, is not a property you inherit from the lending protocol. It is a structuring layer you build on top, and own, and are responsible for. The moment you want a true junior, the question who holds first loss stops being theoretical and becomes a product decision with your name on it. The primitive gives you the safe seat. It does not give you the dangerous one. The dangerous one you have to build, and fill, on purpose.

The Last Unhoused Risk

I have spent this whole essay arguing that risk can be given a native home, and now I have to tell you where the argument stops, because the place it stops is the most honest part of it, and it is the part that keeps this from being a pitch.

Almost everything that bears a cashflow can now be housed. Credit, fixed income, structured products, the vast cashflow-bearing universe that is most of finance: for all of it, the path to a native slot exists, because a cashflow is something you can price, prioritize, tranche, and gate. Where there is a stream, you can build a waterfall.

But there is one risk that still resists, and it is, with a symmetry I do not especially love, exactly the kind of early bet we started with. The seed-stage equity bet has no cashflow. It has no income to separate, no stream to prioritize, no loss waterfall to carve, because there is nothing yet flowing that a waterfall could allocate. You cannot cleanly tranche a thing that does not yet produce. The earliest, most naive-looking bet of all is the one the entire apparatus of this essay cannot yet house. I have made that bet. I will not pretend the machinery I am describing would have given it a home.

And yet. The appetite for tranched early-equity risk is not hypothetical. It already exists, visibly, in the crudest possible form. Liquidation preferences are a two-tier waterfall in a tuxedo: preferred is paid before common, participating preferred double-dips, and the whole seniority stack across rounds is a coarse, hand-negotiated, frozen-in-the-cap-table prototype of exactly the structure I have been describing. The market has been building junior and senior into venture for decades. It has just been doing it badly: bespoke, illiquid, uncomposable, renegotiated from scratch in every deal and then locked in a drawer.

So the frontier is not inventing the demand. The demand has been there the whole time, signing term sheets. The frontier is building the vehicle that the cap table is a rough, illiquid sketch of, the one that would make early-equity risk as legible, priceable, and tradeable as a credit tranche. That vehicle does not exist yet. I am not going to tell you it does. This is the edge of the map, and naming the edge honestly is the only thing that earns you the right to have drawn the rest of it.

The Point

Return, at the end, to the axis. The graded spectrum from principal-protected to first-loss, with all the world's differently-shaped appetite arrayed along it, each pool permitted its own band and forbidden the rest.

The goal was never to make the world risk-loving. The world is already full of appetite for risk, sitting in a thousand shapes, waiting. The goal was never to make anyone braver, because courage was never the constraint. The goal was to build furniture fine enough that conviction could find the capital shaped to bear it, instead of dying in the gap between a risk and the vehicle that could have carried it.

We have spent a long time with one bucket and one word, sorting everyone who arrives early into the same coarse pile and pricing them all as fools. We can stop. We can build the slots, price them, gate them, and let each kind of risk find the capital that was always willing to hold it, if only we had given it a seat with a name.

That is the whole project, and it compresses to a single move: we take risk from naive to native.