Opencurve allocates stablecoin capital to on-chain yield. What distinguishes the book is not where it invests but what it refuses to accept as evidence: no position is opened on a published rate, and every open position is re-priced from chain state daily.
The fund targets a blended real return meaningfully above the on-chain risk-free rate, which is roughly 3.7–4.1% for dollars in a major lending market. We take that spread from three sources: fee income in pools where the arithmetic works, term premium in fixed-rate credit, and curation premium in vaults. We do not take it from leverage, and we do not take directional positions in volatile assets.
Capital is held in separately managed accounts. The fund never takes custody. Each account is a wallet the client controls, on which Opencurve holds a scoped key limited to the venues in the mandate. Clients see the same dashboard the managers use, with the same numbers read from the same blocks.
The venue's published figure is compared against realised share-price growth at 7, 30 and 90 day blocks. A gap larger than half a point ends the review. Roughly two thirds of candidates fail here.
Redeemable liquidity is read directly: idle balance for a vault, available liquidity for a lending market, secondary depth for a principal token. A rate you cannot leave is not a rate.
Pool positions are decomposed to their underlying tokens using live balances. A vault that is 78% weighted into one stablecoin is recorded as that exposure, and sized accordingly.
Each sleeve names the specific thing that would cause a loss, and the rate is judged against that. Where the premium is compensation for illiquidity or credit, we say so rather than describing it as yield.