Transaction Ordering Policy Impacts on Arbitrage

Written by Bessie Liu and Derek Lee

A recently published research paper, whose authors include Akaki Mamageishvili and Ed Felten from Offchain Labs, as well as Robin Fritsch, Maria Inês Silva, and Benjamin Livshits, explores the idea of auctioning off a time advantage for transaction inclusion on Layer 2 (L2s) Ethereum scaling solutions and examines the potential profitability of the arbitrage strategy, compared to other types of transaction ordering policies.

Investigative methodology, simplified

The research paper outlines and compares the potential advantages through arbitrage for a profit-seeking, rational actor using a time-advantage transaction ordering policy with other transaction ordering policies. The authors analyzed the following scenarios in terms of arbitrage:

1. First Come, First Serve (FCFS)

In the first scenario, the FCFS analysis refers to taking an arbitrage opportunity as soon as it appears and calculating total profits. This means that no user has any information advantage outside the market. Users cannot pay a premium for faster inclusion. Instead, users who value fast transaction inclusion are motivated to invest in latency infrastructure to improve their latency to the sequence.

2. Priority Gas Auction (PGA)

In this second scenario, transactions are bundled into blocks created at regular intervals. Traders will compete to be first or earlier in the queue through what is called a Priority Gas Auction or PGA, where traders will add a monetary tip to the block builder, which inflates the amount the traders are willing to pay to get earlier access to the opportunities with “optimal times” within a block. This setup reduces the advantage of fast transaction submissions. Whichever trader offers to pay the highest tip will have a higher likelihood that their transaction will be processed earlier in the block, potentially allowing them to capture arbitrage profit before others.

3. First Come, First Serve (FCFS) with a time advantage actor

In the third scenario, users can place bids ahead of time to gain priority in transaction ordering for a specific period in the future. It is assumed that these users who bid for the fast pass also use a “smart chart” (referred to as dynamic programming) to determine the best moment to execute a trade based on how long the arbitrage opportunity has existed. This smart chart can also tell how much profit is left and how much profit can potentially be made based on current price and time differences. The research paper suggests that the favorable strategy would be to delay trade execution up to the purchased time advantage.

The researchers then ran simulations to see how much profit searchers could make under these three market conditions.

Findings

The research paper shows that the type of transaction ordering policy can influence arbitrage profits. Based on the simulated results, FCFS, with a time advantage, shows favorable results over other transaction ordering policies (from the searcher’s perspective). For a detailed breakdown of the findings, please refer to the research paper.

Breakdown of Research Paper Findings

The research paper suggests that the transaction ordering policy will significantly impact certain MEV extraction operations onchain:

1. The design of transaction sequencing mechanisms significantly affects how arbitrage profits are distributed among participants. According to the research paper, a favorable strategy for time-advantaged arbitrageurs is to delay their trades as long as possible within their time window to maximize profits.
2. An area of ongoing research is how AMM pools can retain and recapture MEV, alongside any downstream effects and implications. Various strategies exist for applications to capture and retain MEV under time-advantaged transaction ordering policies. One approach involves labeling the time-advantaged arbitrageur, allowing AMMs to apply different fees, adjust the market-making function, and even limit the number of transactions the arbitrager can send per interval. This results in a sequential game where the AMM seeks to capture the MEV, and the arbitrageur maximizes their returns. In equilibrium, the paper suggests that if AMM pools attempt to recapture/retain MEV, then the AMM can expect to retain at least 25% of the maximum arbitrage profits while the time-advantaged arbitrager will retain 50%, leaving the remaining 25% unrealized for potential capture.
3. As the research paper shows, FCFS, with Time Advantage, can generate the highest potential profits compared to FCFS (Scenario 1) and PGA (Scenario 2). This is especially true for high-volatility, highly liquid traded pairs like ETH-USDT. The simulation analyzed an ETH-USDT pair in a liquidity pool with $100 million of active liquidity and a 0.05% fee and saw that FCFS with Time Advantage generated profits that were 47.7% more than those of PGA and 86.77% more than FCFS. These figures represent the average minute-by-minute arbitrage profits from CEX-DEX arbitrage, based on data from March 2024, assuming that arbitrageurs are using the FCFS with Time Advantage strategy and the advantage is 200ms (which is the nominal delay proposed in Timeboost.)
4. Negative autocorrelation in price movements can favor FCFS over FCFS with Time Advantage since immediate exploitation of price reversals is beneficial. In the specific simulation mentioned in the research paper (with $100M in active liquidity, 50ms autocorrelation delays, and a constant product AMM), higher negative autocorrelation supposedly leads to greater profits for FCFS arbitrage than FCFS with Time Advantage (FCFS yields 12.79% more). However, FCFS arbitrage and FCFS with Time Advantage remain more profitable than PGA in these conditions, with FCFS with Time Advantage outperforming PGA by 3.67%. Higher profits are associated with more volatile trading pairs (e.g., ETH-USDT) and lower fee pools. In contrast, pairs with lower volatility or higher fees yield less profit under FCFS.