Choosing a Nimiq validator: fees, reliability, and what 0% really means

What a validator fee actually is

Start here, because almost every comparison of Nimiq validators starts in the wrong place. A validator fee is the operator's cut of the rewards it passes on to its stakers. It is not a charge on your stake, it is not taken when you delegate, and it is not a network fee. Staking transactions on Nimiq carry no network fee at all.

Here is the part that matters and that the fee tables never say out loud. The protocol pays the entire batch reward to the validator's reward address, and the split between operator and stakers happens off chain, after that. The chain sees one payment arriving at one address. It does not see, and does not enforce, what happens next.

So a fee is a number attached to a promise about a payment the protocol has no opinion about. That is not a reason to distrust every validator. It is the reason the rest of this page exists, because if the fee were a protocol guarantee you could stop reading after the first column of the table.

Where the rewards come from covers the payment mechanism itself in detail.

Registered, active, elected

These three words get used as if they were synonyms. They are three different states, and only the last one earns anything.

Election happens once per epoch, at the election macro block that closes the previous one, and it hands out 512 slots in proportion to stake, drawn using the chain's random seed. A validator's slot count is what decides its share of the rewards, and it is re-decided every epoch. Being elected last epoch is not a guarantee of being elected in this one.

Some arithmetic that puts the scale in view. On 14 September 2026 we read 52 registered validator records from our own node, 40 of them with no inactivity flag, holding about 9.10 billion NIM between them. Spread over 512 slots, an average slot sits behind roughly 17.8 million NIM. A validator holding a small fraction of that is not assured a slot in any given epoch, and a chooser should expect that rather than be surprised by it.

What happens when a validator goes offline

Nimiq's punishments are aimed at rewards and at eligibility, not at your principal. Worth knowing precisely, because the phrase "you can be slashed" gets imported from other chains and frightens people about the wrong thing.

• A missed block. The reward for that slot is burned and the slot is suspended briefly. A minor, self correcting event.
• Extended time offline. The validator gets deactivated. A deactivated validator is not elected, so it produces nothing and has nothing to distribute, and your delegation sits behind it earning nothing until somebody reactivates the node.
• Equivocation, for example signing two blocks at the same height. The validator is jailed for eight epochs with its rewards burned, and its stakers wait out the jail before they can release funds.

None of those take your delegated NIM. All of them cost you time and earnings, and the last one costs you access as well. That is the real risk profile of picking badly: not loss of principal, but a stake that quietly does nothing for a while. Getting out again is covered in how to unstake NIM.

Payout policy: published, or not

Because distribution is off chain, the payout policy is the product. Two validators can advertise the same fee and behave completely differently.

The official validators list carries three fields worth reading before you delegate:

• Payout type. Reading the live list on 14 September 2026, 19 of the 25 listed validators say they restake, 4 pay out directly, and 2 declare no payout at all. Restaking compounds into your stake; direct sends NIM to your address. Neither is better in the abstract, but they are not the same thing and you should know which one you are choosing.
• Payout schedule. On the same reading, 10 of those 25 published no schedule at all, and read on its own that looks like ten gaps. It is not, and the field definition is why: the schedule is a cron expression belonging to the direct payout type, where it states how often NIM will be sent, and a restake validator has no send to schedule. Of those 10 blanks, 7 are restake and 2 declare no payout. The one that is a real unanswered question is the single direct validator that published no cadence. Twelve of the 19 restake validators publish a schedule string anyway, ranging from "Every 12 hours" to a raw cron expression, which is extra information rather than a required field.
• Fee. 4 of the 25 were listed at 0%, one declared no fee value at all, and the rest sat between 1% and 5%.

A validator that publishes specific, checkable terms is telling you something a validator that leaves a blank where its own payout type needs an answer is not. That is the whole signal, and it is a stronger one than a decimal place of fee. The work is reading a blank field against the rest of the row rather than counting blanks.

Stake concentration, and the choice that affects everyone

Slots follow stake, so delegation decides who produces the network's blocks. On 14 September 2026, across the 52 registered validator records we read:

• the largest single validator held about 8% of all validator balance,
• the largest five held about 36%,
• the largest ten held about 61%.

Inside the narrower official list of 25, which holds roughly 61% of the chain's total validator balance, concentration reads higher still: the top ten of that list held about 84% of the list's stake.

Delegating to whoever is already at the top is the easy click and the one that makes those numbers worse. It is also the one place where a staker's individual choice has an effect beyond their own balance. NimMap draws where the delegation actually sits, if you would rather look at it than take a number for it.

What a 0% fee can and cannot promise

Take it in two halves, because both halves are true.

What it can promise

That the operator intends to pass on everything it receives. When rewards exist, 0% means more of them reach the staker than at 5%. That is real and it is the entire honest case for a low fee.

What it cannot promise

• That rewards exist. 0% of nothing is nothing. An unelected or offline validator at 0% pays exactly as much as an unelected validator at 5%.
• That the protocol is holding it in place. There is no fee field in a Nimiq validator record. The chain stores an address, keys, a reward address, signal data, stake, deposit, staker count and state flags. A fee lives in a list, and a list can be edited.
• That distribution happens at all. The split is off chain. A fee describes the intended split, not the act of sending.

Which is why "0% forever" is worth reading carefully wherever you meet it, including on our own listing, which uses exactly that phrase. It is a commitment made by people who can be held to it, not a constant the protocol enforces. The right question to ask any 0% operator is not whether they mean it today. It is what pays for the server, and what happens to the policy when that answer changes. We answer it for our own validator in why we run ImpactZero at 0% fee, and you should hold every other operator to the same question.

Reading the trust score, including a blank one

The validators list published by Nimiq carries a Validator Trust Score, described in its own repository as a metric to help stakers evaluate performance and reliability. It is built from real block production history, roughly nine months of it: batches a validator was rewarded for, batches it missed, and how dominant it is.

The detail that matters for a chooser is what makes it empty. Those fields are only filled once an epoch has ended, and only if the validator was selected to produce a block. A missing score is therefore usually a statement about history, not about quality. It means: this validator has not been producing blocks, so there is nothing to score.

A worked example, and it is ours

We run a validator on this network, ImpactZero, and on 14 September 2026 it read like this. On chain: registered, no inactivity flag, not jailed, about 3.57 million NIM of total stake, which is our own deposit plus the delegation of 5 stakers. In the official validators list: fee 0%, payout type restake, availability 1, and reliability and total score both empty for epoch 1343.

That last part is the honest reading: we are not being elected. Our stake is a small fraction of an average slot, and until that changes the score has nothing to measure and there are no rewards for a 0% fee to be 0% of. The fair criticisms of us on this page are those: the scale, and the absence of any block production history. Our entry carries no payout schedule, but by the rule above that is the schema rather than a gap, because what we publish is fee 0% and payout type restake, and restake means continuous automatic restaking with nothing further to configure. Why we run ImpactZero at 0% fee sets out the rest of it, including what pays for the server.

We are telling you this rather than hoping you skip the column, because the skill this article is actually trying to hand over is noticing that state on any validator's row. If you are weighing us against the rest, weigh us on that basis. Our validators page lists everyone from the same live data, including the operators whose scores are not blank.

A checklist you can actually run

1. Is it elected, right now? If not, the fee is irrelevant this epoch.
2. Does it have block production history? A populated trust score means the chain has seen it work. A blank one means it has not, and you should ask why.
3. Has it been jailed or deactivated? Both are visible on chain. Zero of the 52 records we read on 14 September 2026 were jailed or retired.
4. What exactly does it publish about payouts? The type first, then the schedule if the type is direct, because that is the case where a blank leaves a real question. Vague is a finding.
5. How large is it already? Concentration is a cost you pay as a user of the network, not just as a staker.
6. Then compare fees. Last, not first, and only among the validators that passed the five questions above.

You are not locked in

Worth knowing before you agonise over the first choice: a validator that stops being elected or stops paying out is a decision you can revise. Nimiq has a dedicated update staker transaction that changes which validator your stake backs without your NIM leaving the staking contract, so there is no withdrawal and no new staker record. It does carry the same lock as unstaking, because the protocol only accepts it once the stake has been deactivated and released, so plan on roughly a day rather than on an instant. The mechanics are in how to stake NIM and the rule itself is in staking mistakes we see as operators.

One constraint to plan around: an address delegates to one validator at a time. Splitting across two validators means using two addresses.

Check all of it yourself

Every figure on this page came from public data you can read without us. Our validators page compares the whole set live, including fees, payout type and trust score. NimMap draws the delegation graph so concentration is a shape rather than a percentage. The block explorer shows the blocks being produced and by whom. If you are earlier in this than choosing, start with what Nimiq is, or browse the Learn hub.

All live figures here are a single reading taken on 14 September 2026 and will have moved since. Nothing on this page is financial advice, and we publish no price figures anywhere on this site.