Tokenized private credit

The instrument is a loan book, and loan books do not quote

A tokenized Treasury holds paper with a live public market behind it. A tokenized private credit vehicle holds loans that were negotiated one at a time and will never trade. There is no closing quote to strike against, so per unit value comes from an amortisation schedule, a delinquency assumption, a loss given default assumption, and a servicer reporting on how the borrowers are actually doing.

Federal Reserve staff make the structural version of this point about tokenization generally: the authoritative valuation stays off chain, and the onchain representation is downstream of it.⁴ For government paper that separation is mostly a latency problem. For a loan book it is a judgement problem, because the number being published was produced by someone with a view rather than read off a screen.

The largest tokenized asset in the category is a home equity book

Figure's HELOC token showed a total asset value of $20,812,600,196 on 3 August 2026, up 6.58% against the prior thirty days, with net asset value at $1.00, an inception date of 17 January 2022, and the asset issued natively on the Provenance blockchain.¹ Read that as a point in time figure on a dashboard that updates continuously, not a fixed constant.

The comparison is what makes it decision changing. The four largest tokenized US Treasury products on the same platform together read $9,562,795,777 the same day.² One home equity loan book carried more than twice the combined tokenized value of the category most people cite when they describe this market. If your mental model of real world assets onchain is Treasury funds, the model is sized to the smaller half.

Par marks hold right up until they do not

A performing loan gets marked at or near par, and it keeps that mark while payments arrive. The mark does not degrade gradually as risk builds, because nothing in the reporting chain observes risk building. It steps down when a payment is missed, a covenant trips, or a servicer reclassifies. Between those events the reported value looks smooth in a way the underlying credit is not.

Maple Finance is the documented case in this market. On 5 December 2022 Maple severed ties with borrower Orthogonal Trading, alleging Orthogonal had misrepresented its financial position, after Orthogonal defaulted on roughly $36 million in loans across Maple pools.³ The loans were reported as performing until they were not. That is the shape of the risk: not a slow slide anyone could trade against, but a level shift landing all at once.

This is also the boundary with the arbitrage terms. In a Treasury fund, a stale reference costs a few basis points until the next strike. In a credit book carried at par through a deteriorating quarter, a stale reference is a mispricing large enough that whoever redeems first is taking value from whoever redeems last.

Redemption promises that outrun the amortisation schedule

Loans repay on their own schedule. A five year amortising loan produces cash monthly and returns principal over five years, and there is no button that accelerates it. So a vehicle offering daily or weekly redemption against a book of multi year loans is funding long assets with short liabilities, which is a maturity mismatch whatever the token contract looks like.

The honest structures say this out loud. They state a lockup, a redemption window, a notice period, a pro rata queue, or a gate with a named trigger, and they size a liquidity sleeve against a stressed redemption scenario rather than an average week. The rest advertise a redemption speed the underlying cannot support and rely on nobody testing it all at once. A redemption promise faster than the liquidation cycle of the asset behind it is a maturity mismatch wearing a feature label.

Where the yield actually comes from

Borrowers paying interest. That is the whole source. Everything else in the stack is a deduction: origination costs, the servicer, the special purpose vehicle, the transfer agent, the platform. The headline rate a holder sees is the borrower rate minus all of it, and the credit spread inside that rate is compensation for the losses that show up in a bad year rather than free money.

So the diligence question is not what the token yields. It is what the borrowers pay, who underwrote them, who services them, what the historical loss rate is on the same collateral type, and how much of the spread the platform keeps. A tokenized credit product needing a token layer incentive on top of the borrower payments is telling you the loan book is not carrying its weight. The token is the distribution rail. The credit is the business.

What we tell issuers to settle first

Five items, none of which live in the contract. Who values the book, on what methodology, and how often that methodology gets independently reviewed. Who services the loans, and what happens to the holders if that servicer fails. What the delinquency and default reporting cadence is, and whether it reaches holders or only the sponsor. The redemption terms, including the gate trigger and who pulls it. And where the loss waterfall puts token holders relative to every other claim on the vehicle.

In our view the valuation review is the most under specified of the five across the structures we see. A sponsor who values the book, reports on the book and controls the redemption terms holds three roles that pull against each other, and separating at least one of them is what makes the reported number worth anything. Whether a specific tokenized credit vehicle is a security, and how its claims rank in an insolvency, is fact specific and jurisdiction specific and belongs to counsel.