Protocol-owned liquidity (POL) is a treasury strategy in which a DAO uses its own capital to provide the liquidity for its token, rather than paying outside liquidity providers to do it. The treasury holds the AMM position itself, so the depth that lets people trade the token belongs to the protocol and cannot be withdrawn by anyone else. It is the treasury-side answer to a structural weakness that dominated early DAO tokenomics: liquidity that was only ever rented, and left as soon as the rent stopped.
The default way to bootstrap a token market in the 2020–2021 cycle was liquidity mining: a protocol emits its own governance token to reward anyone who deposits into its trading pools (Consensys: an alternative to liquidity mining). This buys deep liquidity quickly, but the capital it attracts is mercenary – it chases the highest yield and rotates out the instant a better farm appears or emissions taper. The protocol ends up permanently diluting its holders to rent liquidity it never owns, and a token whose depth depends on emissions is one incentive-cut away from a thin, gappy market.
The critique that named the problem came from the Fei Protocol team, who proposed protocol-controlled value (PCV): if the protocol holds the reserves and LP positions itself, it stops paying rent and starts earning the trading fees instead. Owning the liquidity converts a recurring emissions expense into a durable treasury asset.
Several mechanisms all arrive at the same end – the treasury on the other side of the token's market:
docs.tokemak.xyz now serves the documentation for Auto Finance, a retail yield optimiser. See below.POL is not free, and its most famous exponent became its cautionary tale:
Three of the four routes above are programmes a DAO runs against its own treasury. One was a service it bought, and that market has closed. The closure is easy to miss, because the citations that named it still answer HTTP 200.
Liquidity-as-a-service. The link above resolves, but not to Tokemak. The index served at docs.tokemak.xyz is the documentation for Auto Finance, and it lists Autopools, staking and developer contracts and nothing else: no reactor, no liquidity direction, and no page describing a protocol acquiring depth for its own pair. The introduction names a different audience entirely, framing Autopools around the “complexities” faced by ordinary liquidity providers optimising their own deposits. The token followed the product: the docs record that the legacy TOKE token has migrated to AUTO 1:1, with sTOKE and accTOKE migrating to sAUTO and staking reduced to a single 16-week lock. The pivot predates the rename – the 16 July 2025 capture of tokemak.xyz is already headed “Tokemak Autopilot” and sells automated rebalancing to LPs. Read 17 August 2026.
Bonding-as-a-service. The other way to buy POL was to have someone else run the bond market. Bond Protocol did exactly that, launching on 3 October 2022 with permissionless bonds aimed at treasury growth for crypto projects. Its documentation is still up and still describes those bonds as live. The application is not: on 17 August 2026 neither bondprotocol.finance nor app.bondprotocol.finance publishes an A, AAAA or CNAME record, while the zone’s SOA still answers from Cloudflare – the records were removed, not the domain. Only docs.bondprotocol.finance still resolves, to a GitBook last updated two years ago.
What survives is the in-house half. Olympus still documents its own bonding, and Frax still documents its own AMOs. Both are code the DAO operates against its own treasury, and neither depends on a counterparty staying solvent. That is the durable lesson of the category: a POL strategy bought from a vendor inherits the vendor’s mortality, and depth rented from a shared reserve leaves nothing behind when the reserve winds down. A treasury weighing POL today should ask which of these routes it could still run if every counterparty disappeared.
A caper does not need a separate POL programme, because the liquidity is native to the bonding curve – the contract is the standing counterparty itself. Every buy sends XRD into the caper's own reserve, and that reserve is what honours a sale – the curve is its own market maker, and the depth backing it belongs to the caper, not to an outside LP that can leave (verified against the reserve accounting in contracts/logic/src/lib.rs).
Where Olympus had to acquire protocol-owned liquidity and then defend it, a caper starts with it: the market and its backing are the same object.
That independence is the point of the section above: there is no vendor anywhere in a caper’s liquidity path to wind down, rename, or stop publishing DNS records.