Baseline MEV computation
Note: like with most of our work, the index in this calculation is the validator and the validator operator. MEV profits by searchers and builders will not be the focus.
Post-Merge, validators earn rewards from maximal extractable value (MEV) as well as from transaction fees on Ethereum’s execution layer when they are elected to propose blocks.
MEV extracted from the perspective of the validator is difficult to accurately estimate. Some of it is embedded in priority fess, some is disbursed in direct payments to the validator fee recipient, some of it might be disbursed in a different payment patter between a relayer, a builder and a proposer, and some might not ever print on-chain.
The emergence of out-of-protocol proposer-builder separation (PBS) with mev-boost has introduced a menu of different patterns that make it a non-trivial task to estimate how much MEV was extracted by a validator in a given block.
In this page, we are proposing a case based approach in separating MEV from priority fees on a block by block basis (and then on aggregate) based on information aggregated on-chain and from the various MEV relay APIs.
We are approaching crafting this methodology with an iterative mindset, and have yet to enshrine this to the front-end and API. We want to hear from the community on which approach is the most realistic, representative and fair. Discourse forum [coming soon].
Useful definitions
builder
An agent that constructs a block, drawing from transactions or bundles of transactions sourced from public and/or private mempools.
relay
An entity that gathers blocks and bids from builders. The source of the block that is proposed by a validator who wants to get MEV rewards.
vanilla_block
A block that has not been procured from a MEV relay, and is most likely a build by default spec block.
mev_boost_block
A block that has been procured from a MEV relay.
avg_bid
The average value of bids logged on all relays, in any given block height.
max_bid
The maximum bid observed, from the population of bids received by relays, in any given block height.
winning_bid
The bid that represents the block that made it to the canonical chain as procured from an MEV relay.
fee_recipient
The specified address that receives the priority_fees
in the block constructed by the builder. Depending on the case, the fee recipient could be the address of the builder or the proposer.
priority_fees
Also called block rewards; transaction fees paid to the fee_recipient
by end-users for their transactions to be included in the block.
end_tx
A transaction that takes place at the end of the block and involves the builder that crafted the block and the block proposer. These appear in the Flashbots, bloXroute and Eden relays that prescribe for builders to set the fee_recipient
as their own address and then disburse the reward to the proposer separately, on a subsequent transaction. At the time of writing collectively represent ~90% of the MEV relay landscape.
proposer_account
Address specified by the validator/proposer where they want to be paid rewards from the block by the builder
baseline_mev
The minimum MEV rewards calculated in the block, outside of any priority_fees
MEV value transfer patterns in the wild
There are two main on-chain patterns in which we have observed value transfers between a block builder and a validator; (a) one that involves the end_tx
as parametrized in the table above, and (b) one in which the value transfer happens exclusively via the fee_recipient
.
The following case based approach is based on modulating these two parameters, as well as whether the relay where the block was procured from is known or not.
Case A: Known MEV Relay blocks
Case A.1
This class of cases describe patterns where the builder has set their own address as the fee_recipient
and pays the proposer via an end_tx
(e.g. Flashbots relay).
relay
KNOWN
winning_bid
KNOWN
end_tx
KNOWN
fee_recipient
Builder
If the winning_bid
is greater than or equal to the sum of priority_fees
for the transactions in the block, and , we calculate baseline MEV as:
This is the most straightforward case as the components for computing the baseline_mev
are known. The winning_bid
is captured from the relay and the priority_fees
are computed from the transaction fees in the block.
If the winning_bid
is less than the sum of priority_fees
, for the transactions in the block, we calculate:
In the absence of tools to recreate what the block value would have looked like if the proposer made a block by spec, we classify the winning_bid
they received as the priority_fees
that went to said validator.
Case A.2
This class of cases describe patterns where the builder has set the proposer account as the fee_recipient
(e.g. Manifold relay).
relay
KNOWN
winning_bid
KNOWN
end_tx
n/a
fee_recipient
Proposer
In this case we revert to:
Which implies we book everything that went to validator via priority fees.
While initially looking at this from the winning_bid - priority_fees
lens appears rational, cases like the Manifold bug have shown that bids can be misconfigured to not show exactly how much a validator receives in the end.
Case B: Unknown but likely MEV Relay
This class of cases describe patterns where we don't know the relay the block came from, but given the presence of an end_tx
this is most likely a relay block.
relay
UNKNOWN
winning_bid
UNKNOWN
end_tx
KNOWN
In the absence of data from relays, but with an end_tx
still observed, if the end_tx
is greater than or equal to the sum of priority_fees
for transactions in the block, we calculate baseline MEV as:
Conversely if the end_tx
is less than the sum of priority_fees
for transactions in the block, we calculate baseline MEV as:
We then consider baseline_mev
undetermined and book all the value that went to the proposer as priority_fees
.
Case C: Vanilla block
relay
UNKNOWN
winning_bid
UNKNOWN
end_tx
UNKNOWN
If all of the parameters we are segregating for are UNKNOWN then we revert to :
Qualitatively, if this type of state is observed, it could mean a few things: (1) this is a vanilla block, (2) the bid payment was made out of band (e.g. mev-hiding), or (3) the validator is doing their own block building and MEV extraction.
We believe that these cases in their majority are vanilla blocks, but hold out for emerging patterns as the landscape evolves towards maturity.
Last updated