Paper: arXiv 2310.07865

Authors: Guillermo Angeris, Tarun Chitra, Theo Diamandis, Kshitij Kulkarni

Abstract

Miner extractable value (MEV) refers to any excess value that a transaction validator can realize by manipulating the ordering of transactions. In this work, we introduce a simple theoretical definition of the ‘cost of MEV’, prove some basic properties, and show that the definition is useful via a number of examples. In a variety of settings, this definition is related to the ‘smoothness’ of a function over the symmetric group. From this definition and some basic observations, we recover a number of results from the literature.

Complexity vs Empirical Score

  • Math Complexity: 7.5/10
  • Empirical Rigor: 2.0/10
  • Quadrant: Lab Rats — high math complexity, limited empirical testing

Why this score: The paper uses abstract group theory (symmetric groups), functional analysis (smoothness, spectral properties), and optimization theory to define the cost of MEV, indicating high mathematical density. However, the work is purely theoretical, providing definitions, bounds, and examples without any backtested data, implementation details, or statistical validation, placing it firmly in the theoretical ‘Lab Rats’ category.

Research Flowchart

  flowchart TD
  A["Research Goal:<br>Define 'Cost of MEV' theoretically"] --> B["Methodology:<br>Formalize MEV cost as smoothness of functions over symmetric group"]
  B --> C["Data/Inputs:<br>Transaction ordering manipulations & blockchain data"]
  C --> D["Computational Process:<br>Mathematical proofs & examples analysis"]
  D --> E["Key Findings:<br>1. New MEV cost definition<br>2. Relation to symmetric group smoothness<br>3. Recovery of existing literature results"]

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 formally defines the ‘cost of MEV’ as the excess profit a profit‑maximizing validator can earn from reordering a set of transactions relative to a random ordering.

Source sentence
In this paper, we formally define the ‘cost of MEV’ which quantifies the excess amount that a profit maximizing validator can earn from reordering a set of transactions, relative to a random ordering.

Assumes 5

  • There exists a subset B of transactions that is closed under all permutations (πB = B for all π in the symmetric group). model
    Source sentence
    We will assume that there is a subset B ⊆An closed under permutations, i.e., πB = B for all π ∈S and that f : B →R. If B can be endowed with a metric d : B × B →R+ that is permutation independent, that is, if, for any π ∈S, we have d(π(x), π(y)) = d(x, y), then there are useful sm
  • The value function f on B is permutation‑independent and satisfies a Lipschitz‑like condition |f(x)−f(y)| ≤ L d(x,y) for some metric d that is permutation independent. model
    Source sentence
    We will assume that there is a subset B ⊆An closed under permutations, i.e., πB = B for all π ∈S and that f : B →R. If B can be endowed with a metric d : B × B →R+ that is permutation independent, that is, if, for any π ∈S, we have d(π(x), π(y)) = d(x, y), then there are useful sm
  • The market is modeled as a two‑asset fee‑free constant function market maker with a non‑decreasing concave function G (G(0)=0). market
    Source sentence
    We assume that the market acts like a two-asset fee-free constant function market maker [AAE+22], which we define as a nondecreasing concave function G : R+ →R+ with G(0) = 0.
  • Validators have temporary monopoly over transaction inclusion and ordering in a block. execution
    Source sentence
    In most consensus protocols, a validator has a temporary monopoly over both what transactions are included in a block and the order in which those transactions are executed.
Show 1 more
  • The set of possible actions for frontrunning and sandwiching is defined as described (e.g., A = (R+×{0})∪({0}×R) for actions). model
    Source sentence
    Action space. The set of possible actions in this setting takes the form A = (R+ × {0}) ∪({0} × R) Similar to the previous example, a tuple (x, y) ∈A is an action where x is a trade made by a trader, which is nonzero only when y (the trade made by the adversary) is zero.

Data

No dataset named in the text.

Costs

Not applicable The paper reports no trading or execution result.

Finds 4

  • If the value function f is smooth and the transaction set B has diameter at most t, the global upper bound on the maximum cost of MEV is C_s(f) ≤ 2 L t + inf_{x∈B} C(f,x). theoretical
    Source sentence
    2.1.1 Global upper bounds If we know the function is smooth and the set of transactions B is bounded with diameter at most t, i.e., t ≥sup x,y∈B d(x, y), then we have the immediate upper bound on the maximum cost Cs(f) = sup x∈B C(f, x), given by: Cs(f) ≤2Lt + inf x∈B C(f, x).
  • When the smoothness bound is applied to constant‑function market makers, the bound improves known results and can be saturated (e.g., C_s(f) ≤ 2 L t). theoretical
    Source sentence
    Additionally, if any action x ∈B has xi = xj for every i, j = 1, . . . , n, then the second term in the right-hand-side of (12) is zero, so Cs(f) ≤2Lt. The bound (12) generalizes and improves upon some known results over constant function market makers [CAE22,KDC22], and we show some explicit examples below.
  • As the number of transactions n grows, the cost of MEV approaches 2, saturating the bound provided by the theory. theoretical
    Source sentence
    Taking n ↑∞, we see that the cost of MEV approaches 2, which saturates the bound provided by (10) in the limit.
  • For a constant‑function market maker with total trade volume M, the cost of MEV is at most 8 G′(0) M. theoretical
    Source sentence
    Using the bound (12) and the ‘diameter’ bound (13), we then get that the cost of MEV is at most Cs(f) ≤8G′(0)M, where we used the fact that infx∈B C(f, x) = 0 by setting all traders’ trades to be zero.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 13 related papers). Coverage is partial, so this is neither replication nor refutation.

Cited by 1 paper in the archive (1 by independent authors)

Code

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