Short Paper: Atomic Execution is Not Enough for Arbitrage Profit Extraction in Shared Sequencers

Maria Inês Silva,Benjamin Livshits

2024-10-15

Abstract:There has been a growing interest in shared sequencing solutions, in which transactions for multiple rollups are processed together. Their proponents argue that these solutions allow for better composability and can potentially increase sequencer revenue by enhancing MEV extraction. However, little research has been done on these claims, raising the question of understanding the actual impact of shared sequencing on arbitrage profits, the most common MEV strategy in rollups. To address this, we develop a model to assess arbitrage profits under atomic execution across two Constant Product Market Marker liquidity pools and demonstrate that switching to atomic execution does not always improve profits. We also discuss some scenarios where atomicity may lead to losses, offering insights into why atomic execution may not be enough to convince arbitrageurs and rollups to adopt shared sequencing.

Cryptography and Security

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: **Whether the atomic execution provided by shared sequencers can significantly increase arbitrage profits and thus increase the sequencers' revenue**.

### Background and Motivation

With the development of decentralized finance (DeFi), the trading volume of Layer - 2 expansion solutions such as rollups is increasing. This growth brings more maximum extractable value (MEV) opportunities, and the most common MEV strategy among them is arbitrage. The shared sequencer scheme proposes a new mechanism that allows the transactions of multiple rollups to be processed together, which may enhance MEV extraction and thus increase the sequencers' revenue.

However, at present, there are few studies on the actual impact of shared sequencers on arbitrage profits. Therefore, the authors developed a model to evaluate whether atomic execution will always increase profits when arbitraging between two constant product market maker (CPMM) liquidity pools.

### Main Problems

1. **Does atomic execution always increase arbitrage profits?**
- By constructing a mathematical model, the authors found that switching to atomic execution does not always increase arbitrage profits. In some cases, atomic execution may even lead to losses.

2. **Under what conditions will atomic execution lead to losses?**
- The authors analyzed the influence of different parameters (such as price differences, transaction failure probabilities, etc.) on arbitrage profits and provided explanations in specific scenarios.

3. **Is the design of the shared sequencer attractive enough for arbitrageurs and rollups to adopt?**
- According to the research results, the atomic execution feature alone is not sufficient to persuade arbitrageurs and rollups operators to switch to shared sequencers, because its effect depends on multiple factors, including the probability of transaction failure and external market prices.

### Model Overview

The authors assume that an arbitrageur discovers an arbitrage opportunity for the X - Y token pair on two different rollups. By defining key variables (such as price, reserve, trading volume, etc.), they derived the formulas for the change in arbitrage profit in the non - atomic execution and atomic execution cases:

- **Profit change under non - atomic execution**:
\[
\Delta L_X^{\text{non - atomic}}=\Delta x_B\cdot(1 - F_{S_B})-\Delta x_A\cdot(1 - F_{S_A})
\]
\[
\Delta L_Y^{\text{non - atomic}}=\Delta y_A\cdot(1 - F_{S_A})-\Delta y_B\cdot(1 - F_{S_B})
\]

- **Profit change under atomic execution**:
\[
\Delta L_X^{\text{atomic}} =
\begin{cases}
\Delta x_B-\Delta x_A & \text{if } F_{S_A}=F_{S_B} = 0 \\
0 & \text{otherwise}
\end{cases}
\]
\[
\Delta L_Y^{\text{atomic}} =
\begin{cases}
\Delta y_A-\Delta y_B & \text{if } F_{S_A}=F_{S_B} = 0 \\
0 & \text{otherwise}
\end{cases}
\]

Finally, using the external price \( P_{\text{ext}} \), the total profit can be calculated:
\[
\text{Profit}_i=\Delta L_Y^i+\Delta L_X^i\cdot P_{\text{ext}}
\]
where \( i\in\{\text{non - atomic},\text{atomic}\} \)