Choosing the Best: Using Metrics and Data to Choose the Right Ethereum Validator

Introduction

To begin with, we look at a set of features offered by validators. These are a set of qualitative features that can be difficult to compare, but we believe are important when picking a staking provider.

We then evaluate a set of three indicators that reflect the profitability, security, and reliability aspects of the staking provider. These are a summary of top staking providers by each indicator. We considered only staking providers whose information is publicly available on rated.network, meaning this list isn’t a complete picture of the staking ecosystem. We highlighted providers that are at the top for two or more indicators, which are Allnodes (x2), Attestant (x3), P2P.org (x4), Stakefish (x2).

Qualitative features

Qualitative features are hard to qualify, but they are still important when considering a staking provider. These five features each contribute specific value to the overall quality of a staking provider.

• Self-custody

When a project holds its own Ethereum private keys, it has control over its own funds and reduces the risk of loss due to hacking, theft, or fraud on a third-party platform. When relying on a third party to hold funds, projects are relying on the security measures of that platform, which may not be as secure as it should be. There are numerous historical examples [1, 2, 3, 4, 5]. Self-custody is also a way to preserve a higher level of privacy. Often when using custodial services one is required to provide personal information to the third-party platform holding the funds.

• Audited smart contracts

Since a lot of providers use smart contracts, an audit by independent third-party experts helps ensure the security and reliability of the staking process. The point about using audited smart contracts sounds obvious, but among the top 20 hacks of DeFi protocols, 12 out of 20 were unaudited (including the recent Nomad token bridge exploited for $190M). So it’s important to double-check it.

• Slashing insurance

Slashing is a penalty that can be imposed on stakers who fail to follow the rules of the network or who engage in malicious behaviour, such as double-signing. These events events are rare and slashing in Ethereum is pretty soft. Historical examples show that you can lose up to ±1% of stake in conservative slashing scenario. However, the probability of a disaster-slashing scenario isn’t zero, and testnet showed us that slashing due to bug in a popular consensus client is still a possible option. Such a black swan may lead to dozen percent validators slashed, which increases slashing value due to additional correlation penalty and you can lose up to 1/3 staked ETH.

Slashing insurance is designed to protect stakers against these risks by providing a form of insurance that can help cover losses resulting from slashing penalties. Slashing insurance providers offer coverage to stakers, typically for a fee or a percentage of the staked amount. In exchange, they agree to pay out a portion of the staker’s losses in the event of a slashing penalty.

• Service level agreement (SLA) as a guarantee for future performance

Current performance is not a guarantee for the future, it’s better to have SLA. There have been numerous examples where staking performance changed over time due to operator’s possible outage or decreasing average performance after big network updates.

• MEV-boost enabled

Everyone has the same MEV relays normally, but it’s better to check that your staking provider has all public relays on board since it seriously increases profitability (+1pp. to APR on average). You can check the actual list of public MEV relays here.

Three key indicators for comparing providers

There are three indicators that are the most important when considering staking providers These indicators reflect profitability, security, and reliability.

1. Validator effectiveness as a profitability indicator – This is a better metric than APR because reflects a real provider’s profitability in the long term.
2. Validator’s key management as a security indicator – Bad key management may lead to theft of keys, accidentally or intentionally running a duplicate validator, and slashing (or even losing access to keys and therefore the ability to unstake)
3. Client Diversity as an indicator of network security – Service should not run more than 50% of their aggregate validators with a majority validator client by recommendation of Ethereum.org. That’s because a critical bug in some client soft may lead to a widespread outage or slashing, it’s better to have several clients in this case.

Let’s look at these indicators in detail

Validator effectiveness

The primary metric commonly used for choosing a staking provider is APR. However, due to Ethereum’s complex rewards structure, it’s better to compare staking providers by their validator’s effectiveness rather than APR.

Monthly and even quarterly APRs are highly dependent on a number of random variables like the amount of created blocks or MEV extracted, and it doesn’t reflect the true operator’s performance. A high APR this month doesn’t guarantee a high APR next month.

At the same time, a validator’s effectiveness is a good predictor for APR in the long term, because it measures how well a validator has been performing its duties. So if a validator doesn’t miss attestations and blocks, it has 100% effectiveness and gets 100% of all possible rewards. To illustrate this point, let’s look at the Ethereum rewards structure. How random is the number of blocks created and the block rewards value? How much does it influence APR?

According to the Lido operators’ performance from January 20 to February 20, the average staking APR is 5.7%. However, it will be more accurate to divide it into three separate parts. The split shown below allows us to better analyze the structure of rewards.

1. The first part and largest, we denominate “predictable rewards”. This portion account for just over 50% of the total APR. It represents consensus rewards for attestations, and validators get them every 6.4 minutes for performing attestations. They represent 3.4 pp of the total 5.7 pp of our base APR.

2. The second part of the APR is constituted by random, but predictable rewards, over a long period of time (depending on the number of validators, but one year on average). In total, this portion represents 1.8 pp, and it includes: \

   1. Consensus rewards for the sync committee. Rewards for participating in a group of validators randomly selected every ~27 hours to add their signatures to valid block headers. This allows light clients to keep track of the blockchain’s head without accessing the entire validator set. Although sync committees occur on average every two years, there can be extended periods without selection. If a validator is selected, it is considered an achievement.
   2. Consensus rewards for block proposals. In Ethereum, time is divided into twelve-second units called ‘slots’. In each slot, a single validator is randomly selected to propose a block. Currently, there are 520k active validators, so the chances of being selected are very low. This translates to a validator proposing a block once every 2 months on average.
   3. Execution block rewards with less than 1 ETH (priority transaction fees + an additional fee from MEV) are accrued with block creation as a payment from transactions to the validator for including them in the block. This value is 0.05 ETH on average, but see the picture below, it may significantly vary.
3. The last part accounts for 0.5 pp of the APR. This portion is composed of execution layer block rewards with more than 1 ETH. Like the last part, it is random but unlike the previous one, not predictable because the probability of getting a block reward of more than 1 ETH is less than 1%. When considering the fact that a validator creates 1 block every 2 months on average, the chances of getting these rewards are extremely small.

The chart below showcases the distribution of block value among Lido Node Operators from January 14 to February 14. From this data, we observe that the average block reward is below 0.5 ETH, and the chance of proposing a block with a value higher than 1 ETH is around 0.7%.

Summing up all of the above, we recommend two metrics to evaluate the long-term performance of staking providers: adjusted APR (it’s an APR, excluding every block reward above 1 ETH), or validator effectiveness. Both of these metrics highly correlate with each other, because the validator rewards are limited to a maximum when the validator doesn’t miss attestations and block proposals.

Lido pool operators, December 14, 2022 – February 14, 2023

The last thing related to profitability is the amount of MEV relays. Currently, there are ten active relays. If a validator is connected to every relay, it receives more bids and can pick the most valuable one, thus getting a higher MEV reward. Overall, every staking provider is connected to these 10 MEV relays, so this factor is, at least currently, out of consideration when making a decision. But you can observe above a few outliers with lower APR despite having high effectiveness, which is due to these providers being connected to a limited set of relays.

You can check the top-5 staking providers by validator’s effectiveness and info about the connected MEV relays on rated. At the time of writing, this is the top 5:

Bonus: Forecast your APR!

While understanding the validator’s effectiveness is awesome, everyone wants to forecast their possible rewards and APR. So, P2P.org developed an advanced APR simulator, calculating the APR you can expect on average, considering the growth of total active validators number, possible deviations from the average due to the random amount of block proposals per validator, and random distribution of execution rewards.

Simulate APR and rewards

Validator’s keys storage and management

With the introduction of Proof-of-stake to Ethereum, a new set of cryptographic keys were introduced for validators. These are called “withdrawal keys” and “validator keys”.

The withdrawal key is a key from the withdrawal Ethereum address for unstake and receiving rewards. This address is specified once, and nobody can change it after the staking deposit is sent. Non-custodial staking providers never ask you for the revelation of your withdrawal key. So only you have access to your eth.

Another deal is validator keys. Unlike withdrawal keys, the staking provider has access to validators’ keys needed to sign new blocks and attestations. Therefore your staking provider has to keep them secure. Otherwise, it can lead to theft or loss of keys resulting in

1. Setting up a duplicate validator that leads to slashing
2. Change the fee recipient (execution layer rewards)
3. Losing the ability to unstake

There are several options validators can take to manage these keys. According to Lido operators’@lido/BkxRxAr-o#External-Signers"> statistics there is the following distribution:

• Local storage: Keys stored with a validator (used by 59% of operators*). That’s the most simple variant with pretty low security.
• Remote storage: Keys are stored in a separate machine (used by 19% of operators). That’s the more advanced and secure solution.
• Threshold remote storage: Keys are split into shards and stored on different machines (used by 26% of operators). This is the most robust key management practice.

Note: The sum exceeds 100% because some operators utilize multiple storage solutions. Below we explain the difference between these options, you can dive into the details or jump straight to the list of threshold-powered staking providers

Local storage

To participate as a validator, a key is required to sign blocks and attestations. For this reason, the simplest way to store keys is by storing them together with the validator in the same machine. This is partly because every validator has its own management system which is part of the validator software.

This, however, comes at the cost of security since if an attacker gets access to the machine housing the validator, they also get access to the validator keys and could control the validator or change the reward recipient (execution layer rewards).

Remote storage

For a more advanced solution, the validator and key are stored separately while using a special purpose service called remote signer. This service is responsible for only managing keys and signing data.

All the validator keys are decrypted in a secure environment that is separate from where the signing is performed. As a result, there is no direct access to the keys, and the signing is done only via a special service (Web3Signer, dirk, etc.)

By using a remote storage solution, an attacker can target the validator node, but they can’t get the validator keys from it. They would also need to target a significantly more secure remote signer.

By separating the signing keys from the validator, it is also possible to connect several validators to one signer and reduce the cost of maintenance. However, this storage solution still stores the whole key in one place which can be targeted by an attacker.

Threshold remote storage

In a threshold remote storage setup, the sensitive data is divided into “shares” and each share is encrypted and stored by a separate remote storage provider. To access the data, a requester must retrieve and combine a sufficient number of shares, typically a threshold number, to reconstruct the original data. The use of multiple storage providers and threshold shares provides added security and redundancy compared to a single storage provider.

The use of TRS for Ethereum validator keys can be achieved via Distributed Key Generation (DKG). Distributed Key Generation (DKG) is a cryptographic technique that allows a group of participants to generate a shared cryptographic key without requiring a trusted third party to coordinate the process. In a DKG scheme, each participant generates a unique partial key and shares it with the other participants.

Distributed Key Generation (DKG) can be combined with a k-of-n threshold signature scheme to provide a more secure and decentralized method for managing private keys for validator nodes.

The k-of-n threshold signature scheme requires a minimum of k validators to sign a transaction in order for it to be considered valid, where n is the total number of validators in the group. This scheme ensures that no single validator has complete control over the shared private key, as a minimum of k validators are required to sign a transaction. This provides added security and redundancy, as multiple validators must collaborate to authorize transactions, preventing a single point of failure.

This approach can help improve the security and decentralization of private key management for Ethereum validator nodes, as it allows for the secure and decentralized generation of private keys. Unlike the previous solution, an attacker would need to target different locations to be able to compromise the validator.

Distributed Validator Technology

DVT addresses the limitations found in traditional validator key management by distributing the responsibility among a network of node operators. These operators work together to validate and sign transactions, without the need for a central authority. This approach provides greater security, resilience, and scalability, making it ideal for applications that require a high level of trust and reliability.

The DVT infrastructure is also built using a combination of DKG and k-of-n threshold signing. DKG allows the node operators to generate and distribute cryptographic keys securely, without the need for a trusted third party. k-of-n threshold signing is used to ensure that transactions are only approved when a certain number of node operators have signed off on them. This approach ensures that no single node operator can approve transactions on their own, preventing any one node from becoming a central point of control or failure.

While the potential benefits of DVT are clear, the technology is still in development and not yet widely available. As with any new technology, there are still many challenges that need to be addressed, such as security, interoperability, and ease of use. However, as the technology continues to mature, it has the potential to revolutionize key management and enable new and innovative applications.

Threshold remote signers are used by (alphabet order)

There is limited data about how each operator manages their keys, so we gathered information about threshold-secured operators from public resources like discord and operators’ sites. Therefore this list may be incomplete.

• Attestant
• Cryptomanufaktur
• Certus one
• P2P.org
• Stakely
• Staking facilities

Client Diversity

The Ethereum network is designed to be open and decentralized, meaning it does not rely on any single software client to function. Instead, there are several Ethereum clients developed and maintained by different teams.

Currently, there are several Ethereum clients available, each with its own set of features and characteristics. These clients include Teku, Lighthouse, Prysm, Nimbus, etc. Large validators are also incentivized to use multiple clients, or at least not use the most popular one.

Having multiple clients available for interacting with the Ethereum network is beneficial for several reasons. First, it promotes decentralization by preventing any single client from becoming too dominant and having too much control over the network. Second, it improves security by allowing for multiple implementations of the Ethereum protocol, making it less vulnerable to bugs or attacks on a single client. Third, it encourages innovation by allowing different teams to experiment with different approaches and features.

Ethereum heavily penalizes validators that take part in large-scale outages. And using a single client is an easy way to be vulnerable to an event like that. With proper infrastructure, node operators can quickly move their validators from one client to another if any issues come up.

Client diversity is an important aspect of the Ethereum ecosystem and is actively encouraged by the Ethereum community. Developers are encouraged to use different clients for testing and development, and users are encouraged to use different clients to interact with the network. This helps to ensure the health and longevity of the network and provides a solid foundation for continued growth and innovation.

A list of the clients used by different providers is available on rated. Here is a summary of providers with a diverse client list (alphabet order), this list may be incomplete or outdated, so it’s better to check it by yourself.

• Attestant
• ChainLayer
• Consensus Codefi
• P2P.org
• RockX
• Stakefish
• Staked

The benefits of active decision-making

Staking is a very active part of the Ethereum ecosystem and one that should not be entered into without as much understanding as possible.

The idea behind Ethereum is one of self-ownership. This comes with equal responsibility and as such, putting in the work to understand exactly how to evaluate a staking provider is essential for users looking to participate. We did not set out to rank providers but rather wanted to explore what is important to pay attention to and how to choose a staking provider using numbers and facts. More we clearly understand that any list of providers we mentioned above may be incomplete or outdated. So our last recommendation is to double-check providers on your own (StakingRewards has a list of verified staking providers you can start with).

This article is a part of the Verified Staking Provider Series, which aims to enhance education in the field of staking and validating. To achieve this, StakingRewards invites all Verified Providers to collaborate on research and write articles for publication. Feel free to reach out in our Discord or Telegram.

Disclaimer：

1. This article is reprinted from [[techflowp[staking rewards]. All copyrights belong to the original author [Vladislav Kurenkov]. If there are objections to this reprint, please contact the Gate Learn team, and they will handle it promptly.
2. Liability Disclaimer: The views and opinions expressed in this article are solely those of the author and do not constitute any investment advice.
3. Translations of the article into other languages are done by the Gate Learn team. Unless mentioned, copying, distributing, or plagiarizing the translated articles is prohibited.