How loss is borne

The two pools sit on the same book, but they meet loss in opposite ways. One takes it directly; the other is built to absorb it before it reaches capital. They do not share a loss line, and neither stands in front of the other.

The Single Pool bears its own outcome

A Single Pool is direct, concentrated exposure to one name.

It takes that name's outcome from the first unit, in full. Nothing absorbs loss ahead of it: not the Reserve, not the monetization partner, not another pool, not diversification. Its upside is the whole of what the name returns; its downside is the whole of what the name loses.

Capital in a Single Pool can be lost in part or in full. The structure isolates that risk; it does not remove it.

The Stable Pool is built to hold a baseline

The Stable Pool holds the whole book, and its baseline is supported before its capital is touched.

Two mechanisms stand behind the baseline, in order. First, the book offsets itself: a loss on one name is met by a gain on another, so no single name carries the result. This is protection the book produces on its own.

Second, where the book's realized revenue still falls short of the baseline, a finite support sequence is drawn: the monetization partner's contractual support first, then the Protocol Reserve, to the extent each holds.

Realized revenue, then partner support, then Reserve: a fixed sequence, so the baseline does not depend on any single source holding. It supports the Stable Pool's baseline alone; no part of it stands in front of a Single Pool.

The Reserve

The Protocol Reserve is the last tier of that sequence, and nothing more.

It is finite. It is capitalized from the spread between the book's net return and the Stable baseline, retained rather than distributed, and drawn only to support the baseline when realized revenue and partner support fall short.

It is not a first-loss buffer, it is not held ahead of the pools, and it absorbs nothing for a Single Pool. Once it is exhausted, the Stable baseline can be missed, and capital in any tier can be lost in part or in full.

Scale narrows the band, where the tail permits

As the book grows, the spread of its realized mean narrows. How fast depends on the tail index $\alpha$ (see The thesis).

In the finite-variance regime, $\alpha > 2$, the dispersion of the book's mean falls classically,

for $n$ names in the book: the more names funded, the tighter the band, while the expected return holds.

In the heavy-tailed regime, $1 < \alpha \le 2$, the mean is still well-defined but the variance is not; dispersion falls more slowly, on the order of $n^{\,1/\alpha - 1}$, and a single name can still move a cycle.

The senior tier does not assume the benign regime. Diversification tightens the band where the tail allows; the baseline-support order holds the floor where it does not.

When the support sequence is exhausted

Each tier of the Stable baseline support holds only to the extent it is funded.

If the partner's contractual support is not met, the Reserve is reached earlier, and the counterparty exposure is recorded as such (see Risk).

If the Reserve too is exhausted, the Stable baseline can be missed, and Stable capital is at risk. This bears only on the Stable Pool's baseline; a Single Pool was fully exposed from the first unit, with no tier ever standing in front of it.

No tier is replenished out of a pool's principal, and no support tier stands in front of a Single Pool.

What this protects, and what it does not

Protection here is not the mirror of a shared loss line; the pools have none.

The Single Pool is paid for taking concentration: it carries the full outcome of one name, with nothing behind it, by design. The Stable Pool is protected by the book's own offsetting first and the baseline support sequence second: realized revenue, the partner's support, and the Reserve, none of which is assured.

Neither position can be made to absorb the other's loss. What is fixed is the order in which the Stable baseline is supported, and the separation of the two risk characters.