How will the MEV market size and participant distribution develop in the future?

Author: Leland Lee / Source: galaxy Compilation: vernacular blockchain

Who captures MEV? Who should capture MEV?

Before Ethereum transitioned to Proof of Stake (PoS), Miner Extractable Value (MEV) was primarily captured by Seekers and Miners. Today, on Ethereum, network stakeholders known as builders and validators compete with seekers for MEV revenue. In the future, there will be more participants competing for MEV.

This begs several questions: How is MEV currently distributed among participants? What is the total size of MEV today? What will the overall MEV market and its future distribution look like?

A stylized view of where value goes in a transaction.

MEV is Trading Disease

MEV is the residual value extracted from the blockchain system. As long as there has been a transaction of value through a permissionless system, there has been MEV. For example, if a trader sets the slippage too high, the searcher may get their trades caught in the middle. In this case, the Seeker acquires MEV at the expense of the Trader. To explore the factors affecting MEV, it is necessary to have a taxonomy to understand the growth and decline of the different MEV segments.

High-level taxonomy of MEV.

Although MEV usually has negative connotations, MEV can promote the healthy development of the ecosystem. MEV can create incentives for participants in the system. For example, without seekers to liquidate and arbitrage between venues, the protocol would incur bad debt and users would face wider spreads, making trades get poorer execution on-chain.

Observable MEV is just the tip of the iceberg

While calculating the total addressable market (TAM) of MEV is challenging, we estimate the lower bound for annual MEV extracted from Ethereum to be between $300 million and $900 million, based on estimates from various data providers [1] 。

This MEV TAM is the sum of the value lost by MEV. For example, if a transaction is worth 100 units of value and the user only ends up getting 90 units of value, then 10 units of value are leaked to MEV. In practice, this is calculated by identifying MEV transactions (eg, arbitrage, sandwiches, or liquidations) and then determining searcher revenue. This number can be further broken down into Seeker Profit, Builder Profit, and Validator Net New Funds.

TAM mental model of MEV, not drawn to scale. Source: Flashbots & Leland Lee for illustration.

However, we estimate that the true MEV TAM is not in the $300-$900 million range, and that the 2022 MEV TAM could exceed $1 billion. Calculating the true TAM of MEV is difficult, however, because MEV analysis is best at identifying MEVs that follow predictable or deterministic patterns such as arbitrage, sandwiches, and liquidations. This doesn't capture the long tail of MEV vs. NFT and other DeFi applications, so the actual TAM is actually larger than the estimate. Today, market players are becoming more and more sophisticated in competing with each other. For example, off-chain hedging cannot be tracked, multi-block MEVs are difficult to identify, probabilistic MEVs are almost impossible to detect, etc.

MEV TAM from 2022 to early 2023 according to EigenPhi's observable strategy for arbitrage, liquidation and sandwiches. Note that clearing data only exists after May.

Validators occupy the majority of MEV

The status quo is that validators (also known as proposers or block producers) capture the majority of the observed MEV. Over time, we have seen profitability shift from Seekers to Validators, and more recently back to Seekers. The revenue compression for sandwiches and liquidations is particularly pronounced, as these types of MEV opportunities are highly competitive. Seekers must bid a majority of their expected value to block producers to ensure a transaction is included.

Observable distribution of MEV among the different participants from the merger until the end of April 2023, on Ethereum the distribution is: block producers (56%), searchers (38%), builders (6%) and repeaters (0%). MEV also reached all users of Ethereum through the burning of EIP1559, but this is not calculated.

However, since the total MEV is much larger than the observed MEV, validators may in practice be assigned smaller portions. Additionally, the Unobserved MEV space tends to be less competitive. Therefore, searchers don't have to bid as much for deal inclusion.

Who will capture MEV in the future?

Although both pre-Flashbots and MEVgeth have the same market participants, we suspect that MEVgeth shifts value capture from searchers to miners (block producers). Also, the term "user" is used loosely and can be a user of the dApp's own protocol. May include MEV-burn and other gadgets in the future.

The standard today is three parties capturing MEV - block producers, searchers and builders. However, there is a popular saying that MEV originators (usually transaction originators) should receive rebates for the MEV they generate. MEV-Share (MEV-Share) is the latest instance of this idea, where searchers can bid on order flow from users, wallets and/or dApps. Ultimately, market dynamics drive how much rebates senders receive for the MEV they generate, and how the residual value of MEV is distributed among different market participants. But there is a philosophical question what is the optimal allocation between MEV initiators and searchers and other parties?

MEV aware apps will reduce MEV TAM

However, this debate about fairness and value distribution may be moot. We expect MEV's TAM to decrease due to multiple parties vying for MEV surplus. From an application or dApp perspective, why leak value to searchers when they can keep it for themselves or their users? These "MEV-Aware" applications will implement functions that reduce the MEVs they generate.

This is already happening. For example, MakerDAO and Euler both use auctions for liquidations, as opposed to lenders Compound and Aave, which set liquidations at a fixed discount. In the case of Maker and Euler, liquidation users lose less as custodians compete on price to liquidate loans. In other words, it regains the value they might have lost. For Aave and Compound, there is a fixed liquidation discount. This means that liquidated users suffer fixed losses, rather than variable losses that may be less than a fixed amount. As applications compete for users and liquidity, we can expect dApps to implement features that reduce value leakage to encourage the use of competitors.

How will dApps consider assigning value? The protocol could implement auction fees, transferring some value back to token holders, rather than reducing LP losses through liquidation auctions.

We expect applications to use a better combination of mechanism design (primarily auctions) and smarter parameter defaults to plug MEV leaks that users experience. Some examples include UI warnings to prevent AMM's fat finger, volatility, and pool-aware slippage limits. The famous $2.08 million USDC to $0.05 USDT swap can be easily avoided.

related memes

Looking back at our MEV taxonomy, we expect some classes to be easier to minimize than others. For example, early runs and flanking can be minimized through smarter slippage settings. However, some forms of MEV may be unavoidable – for example, without changes to the underlying agreement [2] , the arbitrage between DEXs cannot be easily reduced.

Who decides the future of MEV?

The MEV market landscape will continue to evolve moving forward and impact various market players. Many questions remain: how will the protocol balance the needs of different user groups, and how will value flow between these groups? Will the market decide how much MEV should be returned, or will the protocol designers intervene? Will users gravitate toward a protocol that better protects them from MEV, or do users simply not care? We expect application-specific MEV to approach zero in the medium term. In the long run, MEV across applications will slowly transfer more value back to the originator of the order flow through protocol design. Ultimately, the protocol and dApp that best protects users will win. Many thanks to Sina, Hasu, Dev Ojha, Walter Smith, and Christine Kim for their conversations and feedback.

Endnotes:

[1] MEV TAM estimates from different data providers.

MEV TAM estimates from various data providers.

[1] Annualized figures for Chorus One are about $92.1 billion per year (their figures start from August 2020 and end with Merge) [2] Chorus One further splits the numbers into good MEV and bad MEV, $713.95 million good (arbitrage and liquidations) + $1206.11 bad (sandwiches) [3] EigenPhi only has data for 2022 and beyond [4] Flashbot does have sandwich data in Dune, unfortunately the data is not particularly clean. However, digging around came up with numbers similar to the Chorus One's top line. [2] Arbitrage between venues is difficult to minimize without changing the underlying L1/L2 protocols. Some designs include auctioning access to the first transaction in a block, but this requires changes to the base layer. While there may be some structure, additional protocol auctions could be conducted if a sufficient percentage of ETH validators restake. [3] Dev from Osmosis pointed out how this can be mitigated by incorporating oracles into Lisk's consensus (forcing ⅓ of the validator set to collude is difficult). Perhaps restaking could be used to mitigate this. Or auction the right to "submit" oracle updates, thus gaining pre/post run rights