Paper: arXiv 2402.06731

Authors: Matthew Willetts, Christian Harrington

Abstract

Convex optimisation has provided a mechanism to determine arbitrage trades on automated market markets (AMMs) since almost their inception. Here we outline generic closed-form solutions for \(N\)-token geometric mean market maker pool arbitrage, that in simulation (with synthetic and historic data) provide better arbitrage opportunities than convex optimisers and is able to capitalise on those opportunities sooner. Furthermore, the intrinsic parallelism of the proposed approach (unlike convex optimisation) offers the ability to scale on GPUs, opening up a new approach to AMM modelling by offering an alternative to numerical-solver-based methods. The lower computational cost of running this new mechanism can also enable on-chain arbitrage bots for multi-asset pools.

Complexity vs Empirical Score

  • Math Complexity: 8.5/10
  • Empirical Rigor: 3.0/10
  • Quadrant: Lab Rats — high math complexity, limited empirical testing

Why this score: The paper presents advanced mathematical derivations for closed-form solutions using Lagrange multipliers and introduces a trade signature concept for N-token pools, indicating high mathematical complexity. However, the empirical validation relies on synthetic simulations and lacks backtest-ready code, datasets, or real-world implementation details, resulting in low empirical rigor.

Research Flowchart

  flowchart TD
  A["Research Goal: Find closed-form solutions<br>for N-token AMM arbitrage"] --> B["Methodology: Derive generic equations<br>for Geometric Mean Market Maker"]
  
  B --> C{"Data / Inputs"}
  C --> D1["Synthetic Data"]
  C --> D2["Historic Market Data"]
  C --> D3["AMM Pool Parameters<br>e.g., constant product k"]
  
  D1 & D2 & D3 --> E["Computational Process:<br>Parallel GPU Execution"]
  E --> F["Comparison: Closed-form vs.<br>Convex Optimization Solvers"]
  
  F --> G{"Key Findings / Outcomes"}
  G --> H1["Better arbitrage opportunities<br>identified"]
  G --> H2["Capitalises on opportunities<br>sooner (faster execution)"]
  G --> H3["Lower computational cost<br>enables on-chain bots"]

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 proposes closed‑form expressions for the optimal arbitrage trade in N‑token geometric mean market maker pools.

Source sentence
We have described the closed-form expressions for the optimal arbitrage trade for N-token pools.

Assumes 4

  • Fees are present and modeled as a multiplicative factor (1‑γ). model
    Source sentence
    With fees (1 −γ), for a trade to be accepted it must be the case that N ∑
  • Each token amount traded is non‑negative and a token cannot be traded for itself. model
    Source sentence
    The pool contains N tokens, so a trader has a vector ∆ of token amounts being traded to the pool for another set of tokens Λ, where entries in each are ≥0. Where ∆i > 0, Λi = 0: tokens cannot be traded for themselves.
  • The trade signature s (indicating which tokens are bought or sold) is known a priori or can be enumerated. execution
    Source sentence
    To use our results, it seems we need a priori the arbitrage trade’s signature s.
  • For low N (N ⪅7) the number of arithmetic operations is reasonable and faster than convex optimisation. other
    Source sentence
    for reasonable numbers of assets (N ⪅7) are substantially quicker to perform than convex optimisation (see Figure 2).

Data

  • synthetic and historic data synthetic unknown
    Source sentence
    in simulation (with synthetic and historic data) provide better arbitrage opportunities than convex optimisers

Costs

Unclear - sources disagree The platform classifier and the paper-text extraction give different answers, so no value is asserted.

Platform classifier: Not modeled

Paper text extraction: Modeled - Fees are modeled as a multiplicative factor (1‑γ) applied to token amounts in the trade invariant.

Source sentence (the authors' words)
With fees (1 −γ), for a trade to be accepted it must be the case that N ∑

Finds 5

  • The closed‑form approach yields better arbitrage opportunities than convex optimisers and can capitalise on them sooner. performance
    Source sentence
    provide better arbitrage opportunities than convex optimisers and is able to capitalise on those opportunities sooner.
  • For reasonable numbers of assets (N ⪅7) the closed‑form calculations are substantially quicker than convex optimisation. performance
    Source sentence
    for reasonable numbers of assets (N ⪅7) are substantially quicker to perform than convex optimisation (see Figure 2).
  • The method scales on GPUs due to intrinsic parallelism, enabling higher speeds than numerical‑solver‑based methods. performance
    Source sentence
    the intrinsic parallelism of the proposed approach (unlike convex optimisation) offers the ability to scale on GPUs
  • In experiments the new approach demonstrated speed and arbitrage performance advantages over current approaches. performance
    Source sentence
    In a range of experiments we have demonstrated the speed and arbitrage performance advantages of this new closed-form approach over current approaches.
  • Even for low‑N pools, enumerating all trade signatures is still considerably faster than convex optimisation. performance
    Source sentence
    for low N-token pool running through all variants is still considerably faster than convex optimisation even prior to any other speedups.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 29 related papers). Coverage is partial, so this is neither replication nor refutation.

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 (18)

  1. Temporal-Function Market Making (2023)
  2. Temporal-Function Market Making Litepaper (2023) — link
  3. QuantAMM Whitepaper (2023)
  4. Uniswap v2 core (2020)
  5. Uniswap v3 core (2021)
  6. KyberSwap: Dynamic Automated Market Making (2021)
  7. Balancer: A non-custodial portfolio manager, liquidity provider, and price sensor. (2019)
  8. When does the tail wag the dog? Curvature and market making (2020) — link
and 10 more
  1. An analysis of Uniswap markets (2021) — link
  2. Liquidity Provider Returns in Geometric Mean Markets (2019) — link
  3. Automated Market Making and Loss-Versus-Rebalancing (2022) — link
  4. Automated Market Making and Loss-Versus-Rebalancing (2021) — link
  5. Ethereum Block Times, MEV, and LP returns (2022) — link
  6. Fast Approximation of the Lambert W Function for Virtual Analog Modelling (2019)
  7. A Sequence of Series for the Lambert W Function (1997) — link
  8. Constant Function Market Makers: Multi-Asset Trades via Convex Optimization (2021) — link
  9. Optimal Routing for Constant Function Market Makers (2022) — link
  10. An Efficient Algorithm for Optimal Routing Through Constant Function Market Makers — → 2302.04938

1 of 18 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.