Paper: arXiv 2404.05803

Authors: Robin Fritsch, Andrea Canidio

Abstract

This paper presents the results of a comprehensive empirical study of losses to arbitrageurs (following the formalization of loss-versus-rebalancing by [“Milionis et al., 2022”]) incurred by liquidity providers on automated market makers (AMMs). We show that those losses exceed the fees earned by liquidity providers across many of the largest AMM liquidity pools (on Uniswap). Remarkably, we also find that the Uniswap v2 pools are more profitable for passive LPs than their Uniswap v3 counterparts. We also investigate how arbitrage losses change with block times. As expected, arbitrage losses decrease when block production is faster. However, the rate of the decline varies significantly across different trading pairs. For instance, when comparing 100ms block times to Ethereum’s current 12-second block times, the decrease in losses to arbitrageurs ranges between 20% to 70%, depending on the specific trading pair.

Complexity vs Empirical Score

  • Math Complexity: 6.0/10
  • Empirical Rigor: 8.5/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper builds upon a formal mathematical framework (Loss-versus-Rebalancing/LVR) requiring discrete-time modeling and numerical simulation, indicating substantial math, while its reliance on extensive historical trade data, Binance price feeds, and simulation of liquidity positions across multiple pools makes it highly data- and implementation-heavy.

Research Flowchart

  flowchart TD
  A["Research Question:<br>Do AMM liquidity providers<br>earn net positive returns?"]
  
  B["Data Collection<br>Uniswap v2/v3 Pool Historical Data<br>Block timestamps & Prices"]
  
  C["Computational Process<br>Calculate Loss-versus-Rebalancing<br>& Compare to LP Fees"]
  
  D{"Analysis Scenarios<br>Arbitrage Loss vs. Block Time<br>Current vs. Faster Block Production"}
  
  E["Key Findings<br>1. Arbitrage losses > Fees<br>2. v2 more profitable than v3<br>3. Faster blocks reduce losses 20-70%"]
  
  A --> B
  B --> C
  C --> D
  D --> E

Evidence Card

What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card

Proposes

The paper empirically measures liquidity providers' losses to arbitrage versus fee earnings across Uniswap v2 and v3 pools and studies how block times affect these losses.

Source sentence
We show that those losses exceed the fees earned by liquidity providers across many of the largest AMM liquidity pools (on Uniswap).

Assumes 6

  • Pools are consistently rebalanced to historical Binance prices. market
    Source sentence
    To simulate arbitrage losses, we assume that the pools are consistently rebalanced to historical prices on Binance, the most liquid cryptocurrency exchange.
  • Blockchain transaction costs are ignored when determining profitable arbitrage opportunities. execution
    Source sentence
    Note that we do not consider blockchain transaction costs when determining profitable arbitrage opportunities.
  • The simulated liquidity position is small relative to the pool size and does not affect trader behavior. model
    Source sentence
    We assume the size of the simulated position to be small compared to the pool size, in the sense that its presence does not affect the behavior of traders and other LPs.
  • During a swap, price stays within a price range with constant liquidity. model
    Source sentence
    This method of fee computation makes an implicit assumption, namely that during a swap, the price stays within a price range with constant liquidity.
Show 2 more
  • Arbitrageurs can trade sufficient amounts at the best bid and ask prices on Binance. market
    Source sentence
    Instead, we implicitly assume arbitrageurs are able to buy and sell sufficient amounts at the best bid and ask prices.
  • Block times are fixed intervals in the simulation (e.g., 100 ms to 16 s). execution
    Source sentence
    We then simulate a hypothetical AMM position in a constant‑product pool for different block times between 100ms and 15s similarly as described in the precious section.

Data

  • Binance second‑by‑second price data · Binance market-prices public
    All tokens traded on Binance that appear in Uniswap pools
    Source sentence
    For all Binance pairs, we retrieve second‑by‑second price data.
  • Ethereum block timestamps · Ethereum blockchain crypto-onchain public
    Ethereum blocks during the observation period
    Source sentence
    Subsequently, at the time of each block (using the timestamps of Ethereum blocks), we consider the external price.
  • Uniswap swap events · Ethereum events crypto-onchain public
    All swaps in the selected Uniswap v2 and v3 pools
    Source sentence
    To compute historical returns from fees, we extract all swap transactions from Ethereum events.

Recognised sources: Binance

Costs

Not modeled A trading or execution result is reported without costs.

Source sentence
Note that we do not consider blockchain transaction costs when determining profitable arbitrage opportunities.

Finds 4

  • Fees earned by liquidity providers are generally smaller than the arbitrage losses they incur, meaning LPs lose money in most large Uniswap pools. performance
    Source sentence
    ... Fees do not sufficiently compensate for arbitrage losses in most of the largest Uniswap liquidity pools, i.e. historically, returns from fees have been smaller than losses to arbitrageur.
  • Uniswap v2 pools are significantly more profitable for passive LPs than the corresponding Uniswap v3 pools for the same token pair and fee tier. performance
    Source sentence
    ... the most‑traded Uniswap v2 pools are significantly more profitable than the corresponding Uniswap v3 pools for the same pair and trading fee.
  • Faster block times reduce arbitrage losses, with 100 ms blocks cutting losses by 20 % to 70 % relative to Ethereum’s 12 s blocks, depending on the pair. performance
    Source sentence
    We find that losses to arbitrageurs are reduced by between 20% and 70% with 100ms block times compared to Ethereum’s current 12 seconds, depending on the trading pair.
  • Empirical loss reduction with block time follows a slower decline than the theoretical square‑root model for intervals shorter than 1 s. theoretical
    Source sentence
    While our empirical findings come close to this model for most pairs and block times larger than 1s, we observe a different regime for block times shorter than 1s. More precisely, arbitrage profits appear to decline more slowly than the theoretical model would suggest.

Checked by others

Supported by

  • Automated Market Making and Arbitrage Profits in the Presence of Fees (2023)
    On the question: How does block time (or block rate) affect arbitrage profitability/losses in AMM markets?
    Compare the findings

    This paper: Faster block times reduce arbitrage losses, with 100 ms blocks cutting losses by 20 % to 70 % relative to Ethereum’s 12 s blocks, depending on the pair.

    That paper: The fraction of profitable blocks (Ptrade) decreases with higher block rates, implying faster blockchains reduce LP losses.

    B shows 100 ms blocks cut losses 20‑70% vs 12 s Ethereum, aligning with A’s result that higher block rates lower the fraction of profitable blocks.

  • Decentralised Finance and Automated Market Making: Predictable Loss and Optimal Liquidity Provision (2023)
    On the question: Do liquidity providers incur net losses due to arbitrage in major Uniswap pools?
    Compare the findings

    This paper: Fees earned by liquidity providers are generally smaller than the arbitrage losses they incur, meaning LPs lose money in most large Uniswap pools.

    That paper: Liquidity providers on average incur a significant loss when providing liquidity in the ETH/USDC pool.

    Both findings report that LPs lose money: A finds a significant loss in the ETH/USDC pool, and B shows fees are smaller than arbitrage losses, causing losses in most large pools.

  • Liquidity provision in CLMMs: evidence from transactions data (2026)
    On the question: Do liquidity providers earn net profit from providing liquidity in AMM/CLMM pools?
    Compare the findings

    This paper: Fees earned by liquidity providers are generally smaller than the arbitrage losses they incur, meaning LPs lose money in most large Uniswap pools.

    That paper: Standalone liquidity provision is profitable only for a minority of LPs, suggesting other motives beyond pool rewards.

    Both findings show LPs generally lose money: A reports fees earned are smaller than arbitrage losses, and B finds only a minority of LPs are profitable.

Qualified by

  • Automated Market Making and Arbitrage Profits in the Presence of Fees (2023)
    On the question: How does block time (or block rate) impact AMM arbitrage profitability and loss magnitude?
    Compare the findings

    This paper: Empirical loss reduction with block time follows a slower decline than the theoretical square‑root model for intervals shorter than 1 s.

    That paper: The fraction of profitable blocks (Ptrade) decreases with higher block rates, implying faster blockchains reduce LP losses.

    Both find profitability drops as blocks get faster; A notes a decrease in profitable‑block fraction, while B reports a slower‑than‑square‑root loss reduction for intervals under 1 s.

Cited by 8 papers in the archive (8 by independent authors): Automated Market Makers: Toward More Profitable Liquidity … (discusses its findings)

Code

No repository by the authors found in the paper.

Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.

Cited References (20)

  1. Automated Market Making and Loss-Versus-Rebalancing (2022) — link
  2. Automated Market Making and Arbitrage Profits in the Presence of Fees — → 2305.14604
  3. Arbitrageurs’ profits, LVR, and sandwich attacks: batch trading as an AMM design response (2023) — link
  4. Batching Trades on Automated Market Makers (2023) — link
  5. Liquidity Provider Returns in Geometric Mean Markets (2021)
  6. The replicating portfolio of a constant product market (2020)
  7. The replicating portfolio of a constant product market with bounded liquidity (2021)
  8. Risks and Returns of Uniswap V3 Liquidity Providers (2023) — link
and 12 more
  1. Behavior of Liquidity Providers in Decentralized Exchanges (2021)
  2. Concentrated Liquidity in Automated Market Makers (2021) — link
  3. Impermanent Loss in Uniswap v3 (2021) — link
  4. Uniswap: A good deal for liquidity providers? (2019) — link
  5. Usage of Markout to Calculate LP Profitability in Uniswap V3 (2022) — link
  6. Uniswap v2 Core (2020)
  7. Uniswap v3 Core (2021)
  8. Let’s run on-chain decentralized exchanges the way we run prediction markets (2016) — link
  9. An Analysis of Uniswap markets (2021)
  10. Diamonds are forever, loss-versus-rebalancing is not (2022)
  11. MEV capturing AMM (McAMM) (2022) — link
  12. Decentralised Finance and Automated Market Making: Predictable Loss and Optimal Liquidity Provision — → 2309.08431

2 of 20 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.