Funding Rate

The funding rate mechanism incentivizes balanced open interest between longs and shorts. The dominant side (more open interest) pays the minority side continuously.

Rate Formula

$$\text{funding\_rate} = \text{r\_funding} \times \frac{|\text{long\_notional} - \text{short\_notional}|}{\text{long\_notional} + \text{short\_notional}}$$

The sign is determined by which side is dominant. When long_notional > short_notional, the rate is positive (longs pay shorts). When short_notional > long_notional, the rate is negative (shorts pay longs). Equal open interest produces a rate of zero.

r_funding is a global parameter in TradingConfig (SCALAR_18) that applies to all markets. The rate is naturally bounded in [-r_funding, +r_funding]. A fully one-sided market produces a rate equal to r_funding, while a perfectly balanced market produces zero.

Accrual

Funding accrues continuously via data.accrue(e, ...), which is called on every market-touching operation (open, close, modify, execute). Borrowing indices are accrued in the same call before funding.

One-sided markets: When either l_notional == 0 or s_notional == 0, funding accrual is skipped entirely even if fund_rate is non-zero. This is because funding is peer-to-peer — there is no counterparty to receive payment. The rate is still SET via apply_funding (reflecting the imbalance), but no indices advance until both sides have open interest.

$$\text{hours\_elapsed} = \frac{\text{seconds\_elapsed} \times \text{SCALAR\_18}}{\text{3600}}$$ $$\text{pay\_delta} = \frac{|\text{funding\_rate}| \times \text{hours\_elapsed}}{\text{SCALAR\_18}}$$

For the paying side, pay_delta is added to that side's funding index (l_fund_idx or s_fund_idx). For the receiving side, the delta is scaled by dominant_notional / minority_notional and subtracted from the index. This scaling factor ensures that total paid equals total received. If longs have 2x the notional of shorts, each short receives 2x the per-unit funding rate. The system is purely peer-to-peer with no protocol cut.

Rate Update Cadence

Lazy Accrual (Every Action)

Every operation that touches market data calls data.accrue(e), which computes funding accrued since last_update, updates both funding indices, and sets last_update = now. This means funding is always up to date when any position action occurs.

Explicit Update (Hourly)

apply_funding() is a permissionless function that enforces a minimum 1-hour interval (ONE_HOUR_SECONDS). For every registered market, it accrues both borrowing and funding indices, then recomputes the funding rate from current open interest balances. It emits ApplyFunding after updating all markets. This ensures rates are recalculated at least hourly, even if no positions are opened or closed.

Per-Position Settlement

When a position is closed, its funding cost or credit is computed as:

$$\text{funding} = \text{notional} \times \frac{\text{current\_index} - \text{entry\_index}}{\text{SCALAR\_18}}$$

Positive funding represents a cost (position paid funding during its lifetime). Negative funding represents a credit (position received funding).

Design Rationale

Continuous accrual (rather than discrete hourly payments) prevents manipulation of the exact accrual timestamp. The dominant/minority scaling makes the mechanism self-balancing, ensuring no value is created or destroyed — funding is purely peer-to-peer with 100% flowing between longs and shorts. LP compensation comes from the separate borrowing fee system, which accrues to the vault via protocol fees. Per-market variation comes naturally from each market's open interest imbalance, even though r_funding is a single global parameter in TradingConfig.