The curve, and the floor beneath it
How x·y=k works, why the last token can never be bought, and where the permanently un-buyable floor comes from.
The equation#
Every ODIN pool is a Uniswap V2 constant-product market. Two reserves — x (ODIN) and y (the paired asset) — hold:
k changes only when liquidity is added or removed — never from a swap. A trade moves reserves along the curve; the fee stays inside the pool, nudging k up.
ODIN's LP tokens are unreachable, so liquidity can never be removed. k has one direction:
Spot price is the reserve ratio, p = y/x. Real size costs more, because the curve bends — to buy Δx ODIN you deposit:
The last token cannot be bought#
As ODIN is bought, the remaining reserve shrinks and each token costs disproportionately more. Push Δx → x and the cost Δy → ∞.
The last token in a pool cannot be bought at any finite price. The reserve is an asymptote.
DeFi slang calls this a "bonding curve." Strictly it's a constant-product AMM curve, not a Bancor supply-mint curve. In ODIN discussion, "bonding curve" means this.
Sixteen automated holders#
Each pool behaves like a non-human market participant that always quotes both sides:
- Always bids — sell into it and it buys, at a defined price, in every market condition.
- Always asks — buy and it sells from reserve, but only along the curve. It rations supply; it never dumps.
- Never deviates — its behaviour was specified in advance and is provable on-chain.
- Deepens —
konly rises, so its depth compounds. A human holder's position does not.
Not smarter than passive holders — immortal, unemotional, and self-deepening.
The floor: k ÷ max supply#
A capped counter-asset can never exceed its maximum supply, so nobody can ever add more than that much y to a pool. The ODIN reserve can never fall below:
Extracting past that point would require more of the paired asset than will ever exist. That residual ODIN is permanently un-extractable — stranded by arithmetic, not merely expensive to reach.
Worked example — ODIN/TSUKA
At a snapshot with roughly 7.21M ODIN against 8.96M TSUKA, k ≈ 6.46 × 10¹³. TSUKA's max supply is 1,000,000,000.
Even if every TSUKA that will ever exist were funnelled into this pool:
Roughly 64,000 ODIN in that pool can never be unlocked, by anyone, at any price.
Reserves drift, so recompute from current reserves rather than reusing these figures — that's the point of showing the method.
And the floor rises. k only ratchets up from fees, so k ÷ max_supply only grows — the one strictly monotonic quantity in the whole picture.
Which pairs qualify#
Ten of the sixteen carry a real cap and therefore a real floor:
SOJ · TSUKA · OHMI · PEAS · WAIT · DBI · APU · APE · SHIB · LINK
Why the other six are excluded
They can still mint, so there's no fixed ceiling to divide by:
- WETH — mints on deposit
- WBTC, OHM, DAI — mint selectors present in bytecode
- PAXG — upgradeable proxy
These give a soft, asymptotic floor rather than an exact one. No cap, no exact number, no line on the ledger.
The sink pool is excluded for a different reason: its ODIN is already counted on the pValhalla line. See burn accounting.
PEAS is the interesting case. It's deflationary, so its cap falls over time — the floor rises on both sides of the fraction: k up, max_supply down.
SOJ is the extreme#
SOJ's entire supply was paired with ODIN from launch. With y capped at the whole supply, the floor is essentially k itself: every fee that lands in the pool becomes stranded ODIN. A one-way absorber. The other capped pairs work identically, just less completely.
Why this matters#
The floor is the strongest single fact ODIN has, because it needs no trust at all. Not a policy, a lock, or a commitment — a quantity of ODIN removed from circulation by two hard-capped supplies and one immutable invariant, computable from public state at any block, with no discretion anywhere in the derivation.
It's also what bounds Liquidity-to-Float from below, and why that ratio has a rising minimum it can never fall through.