Validator Payout Overview
Era Points
For every era (a period of time approximately 6 hours in length in Kusama, and 24 hours in Polkadot), validators are paid proportionally to the amount of era points they have collected. Era points are reward points earned for payable actions like:
issuing validity statements for parachain blocks.
producing a non-uncle block in the Relay Chain.
producing a reference to a previously unreferenced uncle block.
producing a referenced uncle block.
NOTE
An uncle block is a Relay Chain block that is valid in every regard, but which failed to become canonical. This can happen when two or more validators are block producers in a single slot, and the block produced by one validator reaches the next block producer before the others. We call the lagging blocks uncle blocks.
Payments occur at the end of every era.
Era points create a probabilistic component for staking rewards.
If the mean of staking rewards is the average rewards per era, then the variance is the variability from the average staking rewards.
With parachains now on Polkadot, a large percentage of era points will come from parachain validation, as a subset of validators are selected to para-validate for all parachains each epoch, and those para-validators can generate more era points as a result. Para-validators are rewarded 20 era points each for each parachain block that they validate.
In this case, analyzing the expected value of staking rewards will paint a better picture as the weight of era points of validators and para-validators in the reward average are taken into consideration.
HIGH-LEVEL BREAKDOWN OF REWARD VARIANCE
This should only serve as a high-level overview of the probabilistic nature for staking rewards.
Let:
pe
= para-validator era points,ne
= non-para-validator era points,EV
= expected value of staking rewards,
Then, EV(pe)
has more influence on the EV
than EV(ne)
.
Since EV(pe)
has a more weighted probability on the EV
, the increase in variance against the EV
becomes apparent between the different validator pools (aka. validators in the active set and the ones chosen to para-validate).
Also, let:
v
= the variance of staking rewards,p
= number of para-validators,w
= number validators in the active set,e
= era,
Then, v
↑ if w
↑, as this reduces p
: w
, with respect to e
.
Increased v
is expected, and initially keeping p
↓ using the same para-validator set for all parachains ensures availability and approval voting. In addition, despite v
↑ on an e
to e
basis, over time, the amount of rewards each validator receives will equal out based on the continuous selection of para-validators.
There are plans to scale the active para-validation set in the future
Payout Scheme
No matter how much total stake is behind a validator, all validators split the block authoring payout essentially equally. The payout of a specific validator, however, may differ based on era points, as described above. Although there is a probabilistic component to receiving era points, and they may be impacted slightly depending on factors such as network connectivity, well-behaving validators should generally average out to having similar era point totals over a large number of eras.
Validators may also receive "tips" from senders as an incentive to include transactions in their produced blocks. Validators will receive 100% of these tips directly.
Validators will receive staking rewards in the form of the native token of that chain (KSM for Kusama and DOT for Polkadot).
For simplicity, the examples below will assume all validators have the same amount of era points, and received no tips.
Note that this is different than most other Proof-of-Stake systems such as Cosmos. As long as a validator is in the validator set, it will receive the same block reward as every other validator. Validator v1
, who had 18 tokens staked, received the same reward (2 tokens) in this era as v4
who had only 7 tokens staked.
Running Multiple Validators
It is possible for a single entity to run multiple validators. Running multiple validators may provide a better risk/reward ratio. Assuming you have enough DOT, or enough stake nominates your validator, to ensure that your validators remain in the validator set, running multiple validators will result in a higher return than running a single validator.
For the following example, assume you have 18 DOT to stake. For simplicity's sake, we will ignore nominators. Running a single validator, as in the example above, would net you 2 DOT in this era.
Note that while DOT is used as an example, this same formula would apply to KSM when running a validator on Kusama.
Running two validators, and splitting the stake equally, would result in the original validator v4
to be kicked out of the validator set, as only the top v
validators (as measured by stake) are selected to be in the validator set. More important, it would also double the reward that you get from each era.
With enough stake, you could run more than two validators. However, each validator must have enough stake behind it to be in the validator set.
The incentives of the system favor equally-staked validators. This works out to be a dynamic, rather than static, equilibrium. Potential validators will run different numbers of validators and apply different amounts of stake to them as time goes on, and in response to the actions of other validators on the network.
Slashing
Although rewards are paid equally, slashes are relative to a validator's stake. Therefore, if you do have enough DOT to run multiple validators, it is in your best interest to do so. A slash of 30% will, of course, be more DOT for a validator with 18 DOT staked than one with 9 DOT staked.
Running multiple validators does not absolve you of the consequences of misbehavior. Polkadot punishes attacks that appear coordinated more severely than individual attacks. You should not, for example, run multiple validators hosted on the same infrastructure. A proper multi-validator configuration would ensure that they do not fail simultaneously.
Nominators have the incentive to nominate the lowest-staked validator, as this will result in the lowest risk and highest reward. This is due to the fact that while their vulnerability to slashing remains the same (since it is percentage-based), their rewards are higher since they will be a higher proportion of the total stake allocated to that validator.
To clarify this, let us imagine two validators, v1
and v2
. Assume both are in the active set, have commission set to 0%, and are well-behaved. The only difference is that v1
has 90 DOT nominating it and v2
only has 10. If you nominate v1
, it now has 90 + 10 = 100
DOT, and you will get 10% of the staking rewards for the next era. If you nominate v2
, it now has 10 + 10 = 20
DOT nominating it, and you will get 50% of the staking rewards for the next era. In actuality, it would be quite rare to see such a large difference between the stake of validators, but the same principle holds even for smaller differences. If there is a 10% slash of either validator, then you will lose 1 DOT in each case.
CAUTION
If a validator is oversubscribed in an era, staking rewards are distributed only to the the top 256 nominators and the rest of the nominators do not receive any rewards. This is not the case for slashing! Every active nominator of the validator committing slashable offence will be slashed.
Nominators and Validator Payments
Nominated stake allows you to "vote" for validators and share in the rewards (and slashing) without running a validator node yourself. Validators can choose to keep a percentage of the rewards due to their validator to "reimburse" themselves for the cost of running a validator node. Other than that, all rewards are shared based on the stake behind each validator. This includes the stake of the validator itself, plus any stake bonded by nominators.
INFO
Validators set their preference as a percentage of the block reward, not an absolute number of DOT. Polkadot's block reward is based on the total amount at stake, with the reward peaking when the amount staked is at 50% of the total supply. The commission is set as the amount taken by the validator; that is, 0% commission means that the validator does not receive any proportion of the rewards besides that owed to it from self-stake, and 100% commission means that the validator operator gets all rewards and gives none to its nominators.
In the following examples, we can see the results of several different validator payment schemes and split between nominator and validator stake. We will assume a single nominator for each validator. However, there can be numerous nominators for each validator. Rewards are still distributed proportionally - for example, if the total rewards to be given to nominators is 2 DOT, and there are four nominators with equal stake bonded, each will receive 0.5 DOT. Note also that a single nominator may stake different validators.
Each validator in the example has selected a different validator payment (that is, a percentage of the reward set aside directly for the validator before sharing with all bonded stake). The validator's payment percentage (in DOT, although the same calculations work for KSM) is listed in brackets ([]
) next to each validator. Note that since the validator payment is public knowledge, having a low or non-existent validator payment may attract more stake from nominators, since they know they will receive a larger reward.
Last updated