Staking rewards generally derive from a combination of inflationary rewards, transaction fees, and MEV. We are certain that a complete understanding of the Ethereum PBS pipeline allows validators to extract more MEV through targeted infrastructure optimizations.
While adjacent topics have been discussed in literature (e.g. Schwarz-Schilling et al., 2023), Chorus One is the first node operator to successfully test out different optimization approaches on mainnet. We find that the most impactful improvements are contingent on comprehensive internal data, and have generally not been discussed in their specificity.
The goal of this article is to share the results of a recent pilot. We will follow-up with more detail in a later, comprehensive study, which we are co-authoring with one of the most recognizable and competent teams in the MEV space.The pilot makes use of several modifications which positively impact MEV extraction. The rest of the article will discuss one straightforward, illustrative example in more detail, and present the overall results of the pilot so far.
There are two components to APR optimization. Firstly, we should maximize the payoff of the blocks we propose, and secondly, we should minimize the likelihood of missing our chance to propose (i.e. missing our slot). The first of these is more complex and can be approached in several ways, including via latency games, and other infrastructure optimizations.The latter is more accessible, and primarily hinges on running a robust infrastructure setup with appropriate redundancy. However, relay selection also plays a role. Let’s dive in!
A basic illustrative adjustment: drop underperforming relays
The goal of this section is to give an example of a straightforward MEV-adjacent infrastructure adjustment that can positively impact validator APR by minimizing the probability of a missed slot. This modification is not central to our MEV strategy in terms of impact, but it is illustrative of the Ethereum MEV supply chain.
As per conventional wisdom, a validator is best off integrating with a large number of relays, as the mev-boost auction will yield the highest bid, i.e. there is no obvious downside to soliciting a maximum number of bids.
This is only half of the story. Let’s recall how validators and relays interact in more detail. First, the validator requests a block header from a relay, which then delivers the header corresponding to the most profitable block available to the relay. In parallel, the validator also solicits bids from all other relays it is integrated with. The mev-boost auction then determines the highest bid, the validator signs the header associated with this bid, and asks all relays to deliver the payload associated with this header.
The relay that is quickest to respond (typically the relay that delivered the original bid) then broadcasts the block and returns the associated payload to the proposer. This may be done with a delay versus previous implementations (i.e. at proposer’s slot t=0), as early distribution of the payload theoretically allows an unethical proposer to build an alternative block exploiting the transactions in the block received from the relay. This vulnerability has been outlined by the “low carb crusader”, and more details can be found in this post by Flashbot’s Robert McMiller.
The upshot is that in addition to transferring bids from builders, relays also carry responsibility for propagating the final signed block to the network. This is more pronounced now than previously, as the time available for this step has been decreased. Therefore, validators should favor relays that deliver payloads rapidly, or run an idiosyncratic risk of missing their slot, for relays that underperform.In practical terms, we find that relays can diverge significantly on delivery speed, and that for one relay in particular, a routine network disturbance could lead to a missed slot if the validator depends on it to deliver a given block.
The following graph shows the cumulative probability distribution for the maximum time at which a block becomes eligible within each slot, and each line represents a relay.This is a snapshot that is relative to a subset of our nodes over a limited period, and relay performance can vary over time, i.e. should be constantly and granularly monitored. Relays are currently a costly pro-bono good, and we appreciate providers subsidizing the network in this way.
In practical terms, for this particular cluster of nodes, running the relay color-coded light blue is a negative EV decision, i.e. it should be dropped. This is due to the consistent delay it exhibits in making blocks eligible, as compared to other relays.
Our current MEV pilot: A first look at the results
Our current MEV pilot combines straightforward adjustments, like the relay selection process illustrated above, with more significant and systematic infrastructure optimizations.
The below graphs are a first look at the results, and summarize performance over approximately a quarter.
As such, getting a grip on variance requires robust statistical processing. As MEV is tail-heavy (i.e. most profit is produced by rate opportunities), results can vary significantly over time, and capital invested. We are comparing the MEV payoff distribution of the pilot with the MEV payoff realized by a set of Lido nodes. On a per-block level, we find that our pilot has improved MEV rewards significantly:
This extends to the aggregate case - over our sample, the pilot has extracted higher rewards than a “vanilla” setup with a probability approaching 100%:
The upshot is that we feel highly confident that the infrastructure optimizations implemented in our pilot study aggregate out at an APR that is consistently higher than what a non-optimized setup typically achieves.We will elaborate further on specifics in a forthcoming study. If you are interested in learning more about our approach to MEV, please reach out to us anytime at email@example.com.
About Chorus One
Chorus One is one of the biggest institutional staking providers globally operating infrastructure for 45+ Proof-of-Stake networks including Ethereum, Cosmos, Solana, Avalanche, and Near amongst others. Since 2018, we have been at the forefront of the PoS industry and now offer easy enterprise-grade staking solutions, industry-leading research, and also invest in some of the most cutting-edge protocols through Chorus Ventures.