Strategic Liquidity Provision in Uniswap v3

TL;DR

This work addresses strategic liquidity provision in Uniswap v3 by formalizing the dynamic allocation problem where LPs place liquidity in price buckets $B_i=[a_i,b_i]$ and may reallocate with cost $\eta$. It introduces tau-reset dynamic strategies and a neural-network policy (ODRA) to maximize a risk-adjusted utility over contract-market price sequences $\boldsymbol{P}$, leveraging context features and a recurrent optimization framework. Experiments across volatility regimes show that context-aware NN strategies substantially outperform static baselines, with earnings gains modulated by risk aversion and reallocation costs. The results provide actionable guidance for LPs and offer insights into incentives in concentrated-liquidity AMMs like Uniswap v3, informing protocol design and future research directions such as delta-hedging and multi-pool competition.

Abstract

Uniswap v3 is the largest decentralized exchange for digital currencies. A novelty of its design is that it allows a liquidity provider (LP) to allocate liquidity to one or more closed intervals of the price of an asset instead of the full range of possible prices. An LP earns fee rewards proportional to the amount of its liquidity allocation when prices move in this interval. This induces the problem of {\em strategic liquidity provision}: smaller intervals result in higher concentration of liquidity and correspondingly larger fees when the price remains in the interval, but with higher risk as prices may exit the interval leaving the LP with no fee rewards. Although reallocating liquidity to new intervals can mitigate this loss, it comes at a cost, as LPs must expend gas fees to do so. We formalize the dynamic liquidity provision problem and focus on a general class of strategies for which we provide a neural network-based optimization framework for maximizing LP earnings. We model a single LP that faces an exogenous sequence of price changes that arise from arbitrage and non-arbitrage trades in the decentralized exchange. We present experimental results informed by historical price data that demonstrate large improvements in LP earnings over existing allocation strategy baselines. Moreover we provide insight into qualitative differences in optimal LP behaviour in different economic environments.

Figures (15)

• Figure 1: The reserve curve for Uniswap v2. If the pool has reserves $(x,y)$ where $x$ and $y$ represent units of token $A$ and $B$ respectively, then the contract price of token $A$ is $P = y/x$. A trader can send $\Delta x$ units of token $A$ to receive $\Delta y$ units of token $B$, such that $x'y' = L^2$, where $x' = x+\Delta x$ and $y' = y - \Delta y$. The contract price of token $A$ after the trade is $P' = y'/x'$.
• Figure 2: The reserve curve of Uniswap v2 over all prices, and of Uniswap v3 over the price interval $[P_a,P_b]$. When trades give rise to contract prices in this interval, LP assets are swapped according to the v3 curve, which is an affine transformation of the v2 curve and defined to respect the price limits of interval $[P_a,P_b]$.
• Figure 3: An aggregate distribution of liquidity for a Uniswap v3 contract (left plot), with most liquidity allocated close to unit price $P=1$. This results in an aggregate reserve curve (right, red line), which is flatter than the corresponding v2 curve (dotted blue) at prices close to $P=1$ and supports a larger volume of trades at these prices with less slippage.
• Figure 4: Visualization for trade dynamics in an active bucket given by $B_i^* = [a_i^*,b_i^*]$. The blue curve is given by $xy = (L^*)^2$, where $L^*$ is the active liquidity in $B_i^*$. The contract price, $P$, corresponds to the active bundle $(x^*,y^*)$. The upper bound on the amount of token $A$ which can be present in an active bundle for $B_i^*$ is given by $\bar{x}$. Similarly, the upper bound on the amount of token $B$ which can be present in an active bundle for $B_i^*$ is given by $\bar{y}$.
• Figure 5: An illustration of how a $\tau$-reset strategy with $\tau = 1$ can play out. Buckets are represented by circles, and for simplicity we assume that market and contract prices move together at each time step. The shaded circle represents the bucket that contract/market prices are in, and the dynamics of price movements are expressed by the smaller arrows between buckets. Colored buckets represent the contiguous $2\tau + 1$ buckets centered around an epoch's reference bucket. For this sequence, we see that price movements at $t_1 = 2$ and $t_2 = 4$ trigger resets, as the shaded bucket escapes the contiguous $2\tau + 1$ colored buckets. The specific reallocation after each trigger is specified by the allocation function $A$ in the $\tau$-reset strategy.

Theorems & Definitions (14)

• Definition 1: $B_i$-Liquidity Value fan2022differential
• Definition 2: Contract state
• Definition 3: Static liquidity provision strategy
• Theorem 4: Section 3.1 zhao2021understand
• Definition 5: Trading Fee Functions
• Definition 6
• Theorem 7
• Definition 8: Dynamic liquidity provision strategy
• Definition 9: Reset liquidity provision strategy
• Definition 10: $\tau$-reset Strategy