Perspectives · fifteen worked structures
Worked structures.
Fifteen structural problems of a kind that recurs in private markets, each written out in full in the same six parts. They are types, not histories. No manager, counterparty, adviser, asset, value, vintage or date appears in any of them, and none of them would be improved by one.
Each is a room of its own. This page is the register: what each one is, which engagement it exemplifies, and what recurs when the fifteen are read across.
Perspectives · seven rooms
01 · How these are written
The six parts
Fifteen problems, six parts each. No party is named in any of them.
A worked structure is a problem of a kind that recurs, written at the level the problem is actually solved at. It is a class of problem and not an account of one transaction, because the class is the part that transfers to yours.
Six parts, in one order, and the order is the order the work is done in rather than the order it reads best in.
- 01 The structural problem What the manager is trying to build, and the specific thing that makes it difficult. Never the opportunity: the opportunity is the manager's and was settled before anybody arrived. The problem is the distance between the exposure the manager has decided to take and the structures actually available to hold it.
- 02 The relevant constraints What binds, what created it, and the week it surfaces. A constraint written into the fund's own constitution binds differently from one written into a rulebook, because the first can be amended by a vote at a price and the second cannot be amended at all. Sorting them by their source is the first useful thing anybody does.
- 03 The architectures considered Every transaction can be built more than one way. Three routes is the usual count: the route the manager arrived with, the route that becomes obvious once the constraints are stated in one place, and the route nobody had drawn. This is not a menu. Each route does something different to ownership, control, cash flow, administration and exit, and those consequences are the comparison.
- 04 The critical dependencies What has to be true for the preferred route to work, each attributed to the party who can make it true and to the week it has to be true by. A dependency with no owner is not a dependency. It is a hope with a deadline on it.
- 05 The architecture that survives What the constraints leave standing, stated as a structure rather than as a preference: which entity sits where, under which law, what instrument carries the exposure, where governance is written, how cash reaches the register, and which regulatory interface each layer touches.
- 06 The implementation framework What has to happen, in what order, and whose signature closes each item. Written as dependency statements rather than as tasks, because the sequence is the part that is easy to get wrong and expensive to correct. This is the part where the work leaves us.
One class of fact is deliberately absent from all fifteen, and its absence costs the reader nothing.
The identifying facts of a transaction belong to the manager whose transaction it is. What transfers between one transaction and the next is the shape of the problem, and the shape is what is set down here.
A problem written as a type can be held against your own transaction on the same afternoon you read it. A problem written as a history can only be admired.
02 · The register
Fifteen rooms
Ten engagements carry the work. All ten are worked out in full below.
The register runs in the order the rooms are numbered, which is not an order of difficulty. Each line names the problem, the engagement it exemplifies, and the sentence the room opens on. The room itself carries the six parts in full.
A fund that already exists wants an asset that sits under a different body of law. The exposure has been decided.
The manager has found what it wants. The fund exists, and the document that constitutes it does not permit the fund to hold it.
The vehicle is constituted and its tier is fixed. A class of investors regulated under one regime cannot hold it, because their own rules restrict the legal form they may hold, the disclosure they must receive, the oversight the vehicle must carry, or the leverage it may run.
The asset is illiquid, the fund has a term, and the number of parties who could ever buy the asset is small and knowable at entry.
The manager expects to do this kind of transaction repeatedly. Built one at a time, each produces a different chain, a different set of consents, a different process for the administrator, and no reusable answer to any question the last one raised.
The transaction is larger than the fund may take on its own, and the manager wants the balance placed with some of its investors and not others.
The term ends before the value is realised. Selling into the term destroys the thing the fund was built to capture, and holding past it is not something the constitution permits.
A lender agrees to lend against an asset, and the asset cannot be charged where it sits. Either the place it sits creates no interest of that kind over a thing of that kind, or the consent required to grant one belongs to a party who is not in the transaction and gains nothing by being helpful.
A fund opens because one investor commits before the others will. What it is granted to make it commit is a package: a seat, an economic break, a right to be offered co-investment, and a clause saying it will hold whatever anybody else is later given.
The transaction moves an interest somewhere near the top of a structure. Beneath it sit contracts signed years earlier by other people for other reasons, and several of them characterise that movement as a change of control.
The manager holds an architecture drawn by another adviser and wants to know whether it holds before committing resources to it.
The fund wants to distribute an asset rather than money. The administrator cannot value it on the reporting timetable, the register cannot record what a fractional interest would be, some holders are not permitted to receive the asset at all, and the constitution's distribution mechanic assumes cash from the first line to the last.
The vehicle is constituted, the register is populated, an asset is held through a chain beneath it and a facility is drawn against that chain.
A supervised institution commits: a pension scheme, an insurer, a bank, a sovereign body, a fund that is itself regulated.
Nothing has been formed, nothing has been bought and nobody has signed. The house reads that as an absence of structural questions, and it is an absence of only half of them.
Ten engagements carry the work, and the fifteen rooms cover all ten. Which name applies to a transaction is settled by how far it has already gone.
03 · What recurs across the fifteen
The pattern
Most of what binds is already written down. The line that decides is held by somebody who is not in the room.
Read across, these are not fifteen problems. The fifteen rooms name eighty-eight constraints between them. No room names fewer than four, none names more than seven, and ten of the fifteen name exactly six.
Every room sorts them the same way: by source, before severity. Where a constraint was written decides who can move it, how long that takes, and whether moving it is available at all. Five layers recur. The plane below states, for each layer, what it is written in, who can move it, what it costs to establish, and how many of the fifteen it binds.
| Layer · read across | Where it is written | Who can move it | What it costs to establish | Binds in | The rooms |
|---|---|---|---|---|---|
| The fund's own constitution | The document the holders signed, and the allocation, valuation and distribution policies made under it | The register, by a vote, at a price paid in holder confidence | An afternoon with the document | Eleven of the fifteen | 01 · 02 · 04 · 05 · 06 · 07 · 09 · 11 · 12 · 13 · 14 |
| A rulebook, a permission, or the law of a place | Published text held by a body that is not in the transaction, and the register that records title where the asset sits | Nobody in the transaction · unreachable. It moves by application, on the granting body's clock, or it does not move | A question put in the place, and an answer in writing from counsel admitted there | Thirteen of the fifteen | 01 · 03 · 04 · 05 · 06 · 07 · 08 · 10 · 11 · 12 · 13 · 14 · 15 |
| An instrument signed earlier | The facility, the security package, the contracts attaching to the asset, the offer document and the side letters | The party who signed it, by consent, waiver or release, on its own timetable | An hour with the loan documents. The consent takes as long as the party holding it takes | Twelve of the fifteen | 01 · 02 · 04 · 06 · 07 · 08 · 09 · 10 · 11 · 12 · 13 · 14 |
| What an operating party can run | A service agreement, or nowhere at all. It is answered by a person rather than by a rule | The party itself, by system or by appointment, on its timetable rather than the transaction's | One question, put to whoever will carry it every quarter | Seven of the fifteen | 01 · 02 · 05 · 09 · 11 · 14 · 15 |
| What people were told | Nowhere. No document carries it and no adviser holds it | Nobody. It is enforced at the next raise, or on the day the assumption is tested | It cannot be read. It is assembled by asking the people who were in the meeting | Eight of the fifteen | 04 · 07 · 08 · 09 · 11 · 13 · 14 · 15 |
Four observations hold across the fifteen, and not one of them is about the asset.
Eleven of the fifteen carry the same line above their list, and it opens the same way in all eleven: each one, by source. The other four say it in their first sentence. It is not a filing convention. Two constraints in identical words bind differently when one sits in the fund's constitution and the other in a registry's practice, because the first is amendable by a vote and the second is not amendable at all.
Eighty-eight constraints across fifteen rooms. Four at the shortest, seven at the longest, six in ten of the fifteen. The rooms differ in what is on the list and not in how long it runs, which is why a structural problem is finishable rather than open-ended.
Eight rooms name a constraint no instrument holds: what holders were told at subscription, what a credit committee was shown, what an investor's board was told, what the first adviser was asked. In all eight it is the last item on the list. It is enforced by whether the register commits to the next vehicle, by people who owe no explanation for declining.
Nine rooms name a constraint answered on a timetable the transaction does not set: a lender's, a registry's, a granting body's, an administrator's. That item moves the closing date rather than the design, which is why the request goes in the first week rather than the last.
The rooms differ in the asset, the vehicle and the party asking. They do not differ in where the constraint was written, or in what it costs to reach it late.
04 · Where the design stops
Decision ownership
Every engagement names who owns each decision. One of the rows is ours.
The same plane sits behind all fifteen worked structures. Seven decisions, five parties, and one owner per decision. It is drawn at the start of an engagement rather than at the end, because a decision with two owners is a decision nobody takes.
| Decision · owner | The manager | Its counsel | Its tax advisers | Its regulated counterparties | This firm |
|---|---|---|---|---|---|
| Which exposure the fund takes, and whether to proceed at all | The manager | ||||
| Which of the structural routes is adopted | The manager | ||||
| The architecture, the comparison of routes and the stress-test | This firm | ||||
| Whether the structure is lawful in each place it touches | Its counsel | ||||
| The treatment of the chain, the instrument and the exit | Its tax advisers | ||||
| Formation, the documents and the signatures | Its counsel | ||||
| The regulated steps, the administration and the execution | Its regulated counterparties |
Read down the last column, one row of seven is ours. Read across, every other decision has a named owner who is qualified, appointed and answerable for it, and each of them works from the same architecture.
We take a transaction through structural design and execution-readiness, and hand over an architecture in which every dependency is named against the party who owns it. Execution stays with the manager and its appointed counterparties, and the structural analysis is the framework inside which appointed counsel undertakes the relevant validation.
Five worked structures · written as types · stated as at August 2026