Paper Deep Dive: Decentralised Finance and Automated Market Making: Predictable Loss and Optimal Liquidity Provision

Paper: Decentralised Finance and Automated Market Making: Predictable Loss and Optimal Liquidity Provision
Authors: Álvaro Cartea, Fayçal Drissi, Marcello Monga
Date: Forthcoming in SIAM Journal on Financial Mathematics
Estimated Reading Time: 30 minutes

This paper focuses on automated market makers (AMMs) in decentralized finance, particularly constant product markets with concentrated liquidity, like Uniswap v3. It explores the wealth dynamics of liquidity providers (LPs) who strategically adjust their liquidity range to maximize profits. The authors develop a self-financing optimal strategy that balances fee collection, predictable losses (PL), and concentration risk. Additionally, they examine the performance of LPs using Uniswap v3 data for the ETH/USDC pool. The study finds that most LPs have incurred significant losses historically, while the proposed optimal strategy outperforms typical LPs by leveraging market dynamics and volatility.

Core Insights:

1. LP Wealth Dynamics: The wealth of LPs in AMMs consists of fee income, the value of holdings in the liquidity pool, and rebalancing costs. Strategic LPs dynamically adjust their liquidity ranges to optimize these components.
2. Optimal Strategy: A closed-form optimal liquidity provision strategy is derived. It takes into account profitability (fees minus gas fees), predictable losses (due to holding assets in the pool), and concentration risk.
3. Concentration Risk: Concentrated liquidity increases potential fee revenue when the asset's exchange rate is within the LP's specified range. However, it also introduces the risk of not collecting fees when the exchange rate moves outside this range.
4. Volatility Impact: Higher market volatility incentivizes LPs to widen their liquidity range to minimize PL. The strategy suggests that when volatility is extremely high, withdrawing from the pool might be optimal.
5. Historical Performance: Analysis using Uniswap v3 data shows that LPs often trade at a loss. The proposed strategy's out-of-sample performance surpasses the historical returns of the typical LPs in the analyzed ETH/USDC pool.

The paper thoroughly examines how decentralized finance (DeFi) platforms and automated market makers (AMMs), particularly Uniswap v3's concentrated liquidity pools, affect liquidity providers (LPs). The primary focus is on developing an optimal liquidity provision strategy that considers fee income, predictable losses (PL), and concentration risk. The authors model the continuous-time wealth dynamics of strategic LPs who frequently adjust the range within which they provide liquidity to optimize profitability.

The authors begin by describing the mechanics of constant product markets (CPMs) with concentrated liquidity (CL), emphasizing their widespread use in decentralized finance, notably in platforms like Uniswap v3. In contrast to traditional market structures like limit order books, these AMMs use liquidity pools where LPs deposit assets, and liquidity takers (LTs) trade directly with the pool. LPs in CPMs with CL must choose specific intervals, or ranges, for their liquidity, directly influencing their potential fee revenue and risk exposure.

Liquidity Provider Wealth Dynamics

The paper dissects LP wealth into three main components:

1. Position Value: This refers to the LP's asset holdings in the pool. As market conditions change, LPs may incur predictable losses due to changes in the asset values they have deposited.
2. Fee Income: LPs earn fees proportional to the liquidity they provide, which is directly influenced by the range of asset prices they select. Concentrated liquidity strategies can increase fee revenue when the market price remains within the LP's chosen range but simultaneously heighten risk.
3. Rebalancing Costs: Adjusting liquidity ranges incurs costs, especially when the LP needs to rebalance asset holdings, which can involve trading fees or gas fees on the blockchain. The authors model these costs as proportional to the amount of rebalanced assets.

Optimal Liquidity Provision Strategy

The strategy proposed by the authors considers multiple factors, including the profitability of the liquidity pool (derived from fee income and rebalancing costs), PL, and concentration risk. The PL concept introduced here quantifies the losses LPs cannot hedge, stemming from both the depreciation of assets in the pool (convexity costs) and the opportunity cost of locking assets. The study reveals that as LPs narrow their liquidity range, PL increases, suggesting that tight concentration can lead to higher potential losses. The authors construct a self-financing strategy based on the maximization of an LP's terminal wealth, incorporating the following:

• Profitability: The LP's range width is adjusted dynamically based on the pool's profitability, which considers both provision fees and incurred gas fees.
• Concentration Risk: Defined as the risk of the LP's range no longer capturing fee revenue when the exchange rate exits the specified range. When market volatility rises, the strategy adjusts by skewing the liquidity range or widening it to mitigate concentration risk.
• Stochastic Drift: The model accommodates scenarios where the exchange rate's drift is stochastic, allowing the strategy to optimally skew the range to capture expected market movements and maximize fee collection.

The paper provides a closed-form solution for the optimal liquidity range. The results indicate that the LP's decision to narrow or widen their liquidity range depends on market volatility and fee profitability. In cases of high volatility, the optimal strategy suggests widening the range to lower exposure to PL, while in more stable conditions, LPs can benefit from concentrating liquidity around the current market rate to increase fee revenue.

Empirical Analysis Using Uniswap v3 Data

The study uses historical data from Uniswap v3's ETH/USDC pool, spanning from May 2021 to August 2022, to validate the model. The analysis highlights that many LPs historically suffer significant losses, primarily due to inappropriate range selections and market conditions moving beyond their concentrated liquidity ranges. Key observations from the empirical analysis include:

• Average LP Losses: The data suggests that most LPs have experienced substantial losses on average, which the authors attribute to market dynamics and insufficiently optimized liquidity ranges.
• Superiority of Proposed Strategy: By applying the optimal strategy to historical data, the study demonstrates its superior out-of-sample performance compared to the average returns of LPs in the pool. The strategy manages to adapt to changes in market volatility and fee rates, resulting in more consistent fee income and reduced PL.

Market Implications and Further Considerations

The study’s findings have several implications for the tokenomics of decentralized finance:

• Fee Structures: The optimal strategy’s dependence on pool profitability underscores the need to carefully consider the fee structures in AMMs. Higher fees attract liquidity but can also exacerbate PL if not paired with proper liquidity range management.
• Market Volatility: The relationship between market volatility and optimal liquidity provision suggests that LPs must be vigilant about adjusting their strategies as market conditions shift. In highly volatile environments, the risk of PL increases, urging LPs to adopt wider liquidity ranges or even withdraw entirely.
• Concentration Risk Management: The concept of concentration risk introduces the need for strategies that can dynamically adjust to ensure that liquidity remains in profit-generating ranges. The study suggests that there is a trade-off between concentrating liquidity (to earn higher fees) and risking losses if market prices move out of the range.

Questions for Further Exploration

1. Optimal Fee Structures: How might AMMs redesign their fee structures to mitigate the predictable losses faced by LPs while maintaining market liquidity? Could a dynamic fee model be introduced to reflect changes in market conditions?
2. Signal-Based Adjustments: Can LPs leverage additional predictive signals beyond exchange rate drift to inform liquidity range adjustments and further optimize their profitability?
3. Mitigating Gas Fees: Since rebalancing costs and gas fees are significant factors affecting LP profitability, what developments in blockchain technology (e.g., layer-2 solutions) could help reduce these fees, thereby enhancing overall liquidity provision efficiency?
4. Impact of Pool Depth: How does the depth of liquidity pools influence individual LP strategies, particularly when multiple LPs use similar optimization methods? Could this lead to changes in market dynamics and fee revenue distribution?