Skip to content
log(swap)

Liquidity and LVR

Gamma is liquidity

What a trade prices against is the curvature of the position. The liquidity a swap meets at pp is the dollar gamma

λ(p)  =  p2 ∣V′′(p)∣\lambda(p) \;=\; p^2\,\lvert V''(p)\rvert

in quote units — Uniswap v3's liquidity parameter is exactly this quantity. More gamma per dollar of position is more fill per dollar of position.

For the log position, gamma is constant. By I3, λ=L\lambda = L: the same liquidity at every price. A single parameter therefore sets the pool's whole liquidity exposure — LL is the liquidity, with no price attached — and a bp-per-dollar quote stays true as price moves.

And it is higher per unit of capital

With V=L(1+ℓ)V = L(1+\ell),

λV=11+ℓagainstλV≡14  for xy=k\frac{\lambda}{V} = \frac{1}{1+\ell} \qquad\text{against}\qquad \frac{\lambda}{V} \equiv \frac14 \ \ \text{for } xy = k

— for constant product V=2kpV = 2\sqrt{kp} and p2∣V′′∣=V/4p^2\lvert V''\rvert = V/4 at every price. So liquidity per unit of capital is up to 4×4\times constant product, exactly 4/(1+ℓ)4/(1+\ell), maximal at the floor.

The reason is not a better curve. It is that the position holds nothing below its floor: all of its capital works where price is, instead of being spread over (0,∞)(0,\infty). Concentration comes for free, from the boundary rather than from a fee-tier or a range-management layer.

The bill for liquidity is LVR, and it is the same λ\lambda

Under geometric Brownian motion with volatility σ\sigma, loss-versus-rebalancing runs at

LVR(p)  =  12σ2p2∣V′′(p)∣  =  12σ2λ\mathrm{LVR}(p) \;=\; \tfrac12 \sigma^2 p^2 \lvert V''(p)\rvert \;=\; \tfrac12\sigma^2 \lambda

for every invariant. No curve buys liquidity at a discount — the 4×4\times above costs exactly nothing per unit of liquidity, and everything per unit of capital saved. For the log position

LVR=12 σ2L,\mathrm{LVR} = \tfrac12\,\sigma^2 L,

independent of price, and unchanged by the cap.

Why constant matters

The revenue needed to break even against informed flow is the same wherever price sits. In Uniswap v2 the same rate scales as p\sqrt p, so the adverse-selection bill depends on where price happens to be — a pool calibrated at one price is miscalibrated at another, with no trade having gone wrong.

Constant LVR is what makes a single global fee calibration meaningful. It is the precondition for everything in Fees as a path statistic: one ϕ\phi compensates uniformly, and the realized edge is a subtraction rather than a model.

The units, together

quantityvaluedepends on price?
liquidity λ\lambdaLLno
LVR rate12σ2L\tfrac12\sigma^2 Lno
fee rateϕL∣dx∣\phi L \lvert dx\rvertno
position value VVL(1+ℓ)L(1+\ell)yes, through ℓ\ell
liquidity per unit capital1/(1+ℓ)1/(1+\ell)yes, through ℓ\ell

LL sets risk and reward together, and it is the only number that does. The floor sets how much capital is standing behind that same LL — which is why the two knobs an LP is offered are orthogonal.

Scale-covariance

V′′V'' is homogeneous of degree −2-2, so p↦λpp \mapsto \lambda p acts as (Q,R)↦(Q+Llog⁡λ,  R/λ)(Q, R) \mapsto (Q + L\log\lambda,\; R/\lambda) and preserves the form. Floors enter only through x−ξx - \xi. The pool has no privileged price scale, and the only thing that distinguishes one market from another is where its floors sit relative to spot.