Background
There has been a large amount of high quality research done on MEV auction, including empirical study, game theory and mechanism design. This research raised issues across Ethereum’s Proposer-Builder-Separation (PBS) stack:
- Validator layer behaviors (timing game, missed slots, validator monopoly)
- Builder layer behaviors (centralization risk, vertical integration advantage, orderflow composition, and strategic biddings)
However, there are a few key questions about MEV remain unanswered:
- How much profit do non-atomic (or CEX-DEX) MEV searchers make on ETH L1?
- How much MEV occurs on L2s?
We'll explore these topics in detail. Additionally, with changes to the PBS stack and the Ethereum protocol layer, new market developments arise:
- How will relay market dynamics change with profitable relays?
- How have blobs affected PBS auctions?
Problem Statement
ETH L1 Track
Measuring Profitability and Advantage of non-atomic MEV on Block Bidding
Currently, two vertically integrated searcher-builders engaged in non-atomic CEX-DEX arbitrage account for up to 70% of Ethereum block production. Their monopolistic positions as builders are driven by arbitrage profits. Understanding the scope of these profits can help us assess the advantage this provides the builder in winning the block auction. Additionally, identifying patterns in their PnL trends within a slot can demonstrate the structural advantage gained from their dual role as a searcher-builder, which allows them to control the finality of the on-chain trade.
Profit estimation on CEX trades has been challenging due to their offchain and anonymous nature. Here's a framework to guide the analysis:
- Build heuristics to identify CEX-DEX MEV arbitrage on DEX and quantify its volume.
- Estimate PnL using CEX data price feeds and markouts.
- Compare searchers’ PnL across their different Block Building profiles, and Market Making profiles across RFQ platforms.
- Measure the structural advantage a builder role provides for non-atomic MEV's PnL.
Gas and Block Auction Overpayment
On Ethereum, users compete in gas auctions to land their transactions without knowledge of other incoming transactions. Builders also compete in block bidding auctions without knowing the true value of competing builders' blocks, raising their bids based on the highest bid posted by relay and even adding subsidy. This overpayment can lead to higher costs for users and builders. How can we quantify and address this inefficiency to improve the design and avoid overpaying validators? We can explore this from various perspectives:
- Overpayment of MEV Value: Assessing the total MEV value from all submitted orderflow, including gas priority fees and coinbase transfer value.
- Relay Captured Profits: Estimating the surplus of the delivered bids historically compared to a second-price auction; which equivalently is to measure the profits (or "kickbacks") of a performant relay, using UltraSound's bid adjustment feature as a case study.
- In relation to this topic, evaluate the potential impact of a competitive builder launching a relay and capturing its value.
Blob Fee Market and its implication on PBS
Post-EIP4844, we aim to understand:
- How blobs have affected the block-building game and how severe the contention issue is.
- How to price blobs more efficiently.
- How latency increases with data size for builders and relays.
- How to make blob fee mechanisms resistant to spamming (e.g. inscriptions).
Redesigning PBS
There have been a few proposals thinking about possible alternate market designs instead of PBS: these include PEPC-Boost, Execution Tickets and others. Overall the goal is to reduce the centralization pressure on the builder portion of the stack, potentially increasing the efficiency of proposed blocks and decreasing the fees paid (see above). Another goal is to reduce inefficiencies from double-marginalization (if, e.g., increasing amounts of order flow are being sold privately in the form of OFAs etc), then how do these upstream auctions interact with the downstream PBS auction? (see e.g. this paper in a different context). Possible questions include:
- How do these designs compare in the context of a formal game theoretical model on things like revenue and efficiency?
- Are there alternate designs the community should be investigating further (or even within current designs, for example could a sealed-bid MEV boost auction be better than the current ascending auction?)
L2 Rollups Track
Quantifying L2 MEV and potential impact of PBS implementation on sequencer design
The Dencun upgrade has significantly reduced L1 batch submission fees, attracting more activity to L2. The single-sequencer infrastructure setup of L2, coupled with its lack of an intrinsic mempool, limits MEV strategies to statistical arbitrages and spamming tactics. This has led to congestion on L2 networks, with Priority Gas Auctions (PGAs) in play — for example, Optimism and Base — resembling Ethereum's pre-Flashbots state.
It's crucial to gain a deeper understanding of MEV activities on L2s. However, so far, no agreed-upon numbers exist for L2 MEV, validated through multiple sources and methodologies. Moreover, there's a lack of live monitoring data - similar to works done for Ethereum (e.g. mev-inspect, libmev, eigenphi) - for L2 MEV volume and searcher profits. Ongoing efforts by different teams are being made, and we encourage further work along these lines to be pursued under the scope of this grant.
Some L2 MEV datasets and research published so far include:
- open source dataset built by hildobby on Dune Analytics (heuristic links: Sandwiched | Sandwiches | Atomic Arb)
- A paper (by Arthur & Luca) quantifying Polygon, Optimism, Arbitrum MEV using a mev-inspect implementation (research grant by Flashbots)
Beyond accounting for MEV volume and profits, L2 MEV strategies and compositions might differ from ETH L1 due to L2's sequencing rules. Additionally, there are new design challenges and questions for L2 to explore MEV democratization solutions, or to adopt an ETH L1 PBS-like setup, despite its issues of builder-centralization and validator-monopoly.
Places to Start
