Non-Atomic Arbitrage in Decentralized Finance

TL;DR

This paper reveals that non-atomic arbitrage—price discrepancies between on-chain DEXes and off-chain markets—accounts for a substantial portion of Ethereum DeFi activity during the PoS era. By combining on-chain DEX data, PBS relay activity, network observations, and off-chain price data, the study demonstrates that over a quarter of volume on Ethereum's top five DEXes can be attributed to non-atomic arbitrage, with eleven searchers responsible for more than 80% of this activity. It also uncovers strong centralization signals in the PBS block-building market, driven by high-volatility periods when specialized HFT builders dominate block selection. The work discusses security risks, such as potential time-bandit-like behavior and increased block congestion, and proposes mitigations like separating top-of-block extraction and adjusting block times to reduce the profitability of such arbitrage.

Abstract

The prevalence of maximal extractable value (MEV) in the Ethereum ecosystem has led to a characterization of the latter as a dark forest. Studies of MEV have thus far largely been restricted to purely on-chain MEV, i.e., sandwich attacks, cyclic arbitrage, and liquidations. In this work, we shed light on the prevalence of non-atomic arbitrage on decentralized exchanges (DEXes) on the Ethereum blockchain. Importantly, non-atomic arbitrage exploits price differences between DEXes on the Ethereum blockchain as well as exchanges outside the Ethereum blockchain (i.e., centralized exchanges or DEXes on other blockchains). Thus, non-atomic arbitrage is a type of MEV that involves actions on and off the Ethereum blockchain. In our study of non-atomic arbitrage, we uncover that more than a fourth of the volume on Ethereum's biggest five DEXes from the merge until 31 October 2023 can likely be attributed to this type of MEV. We further highlight that only eleven searchers are responsible for more than 80% of the identified non-atomic arbitrage volume sitting at a staggering $132 billion and draw a connection between the centralization of the block construction market and non-atomic arbitrage. Finally, we discuss the security implications of these high-value transactions that account for more than 10% of Ethereum's total block value and outline possible mitigations.

Figures (17)

• Figure 1: PBS scheme visualization. In step ①, a searcher sends (bundles of) transactions to one or many builders privately, the transactions included in the bundle can be from the public Ethereum mempool, private order flow or from the searcher itself. The builder then builds a high-value block with bundles received from searchers, as well as transactions from the public mempool or private order flow. In step ②, the builder sends the block to the relay. The relay checks that the block complies with its policies. Then, if requested by the proposer, the relay passes the highest valid bid and corresponding block header on to the proposer in step ③. The proposer chooses the highest value block amongst the blocks received from the relays, signs the header, and returns it to the relay. This prompts the relay to reveal the full block and broadcast it to the public Ethereum network in step ④.
• Figure 2: Non-atomic arbitrage illustration. A price difference between DEXes and off-chain markets occurs (step ①), imagine that the ETH-USDT price is higher on off-chain markets than on-chain markets. In step ②, the searcher submits transaction $T_A$ to profit from this price difference to the builder, i.e., buys ETH for USDT on DEXes. The builder then includes transaction $T_A$ (as long as the fees paid by the transaction are sufficient) in the block and forwards the block to the relay along with the bid (step ③). The relay chooses the block with the highest bid to pass on the to proposer (step ④). From all relays, the proposer picks the highest value block, and if this block includes $T_A$ the arbitrage trade executes. Importantly, this arbitrage opportunity is non-atomic as it includes swaps on- and off-chain.
• Figure 3: Arbitrageur profit as a function of the price difference for $L =10^6$, $P_\mathrm{on}=1$, $f=0.3\%$, and $g=0.1\%$. Note we only plot the profit in the profitable range.
• Figure 4: Bids submitted by builders for block 18,360,789 on 16 October 2023, along with the ETH-USDT and BTC-USDT price on Binance.com. Notice that the bids, especially those previously identified "HFT" builders Gupta2023Centralizing start to rise significantly shortly after the prices on Binance.com start to move. Importantly, this price movement creates an arbitrage opportunity between DEXes on Ethereum and Binance.com. Further, the bids submitted by HFT builders exceed those by the remaining builders significantly.
• Figure 5: Cumulative volume of non-atomic arbitrage swaps by searcher. In total, the non-atomic arbitrage volume during our data collection period is $132 billion. We provide a mapping from the searcher address to the searcher name as part of our data set anonymous2023.