Choosing a validator is choosing an operator, not picking a number on a list. When you delegate, your stake rides on somebody else's server, somebody else's upgrade habits, and somebody else's response at 3am when something breaks. The right choice is a validator with a fair commission, a long record of staying online, enough of its own stake at risk to care, and a public history of honest communication. This post shows you how to check all four before you delegate a single token.
If the words validator and delegator are new to you, read our validator vs delegator guide first, then come back. If you already understand the roles, the checklist below takes about five minutes per validator, and it is worth doing for every delegation, not just the first.
Commission: what it is and what it is not
Commission is the share of staking rewards the validator keeps for running the node. It is usually shown as a percentage on the chain's explorer or staking dashboard. A validator charging 5 percent keeps 5 percent of the rewards your stake earns and passes the rest to you. Commission rates vary widely by chain and by validator, so there is no universal good number: check the current rates on the chain's explorer on the day you delegate, and compare validators on the same chain, not across chains.
The trap is treating the lowest commission as the best deal. A validator charging zero commission earns nothing for running the node, which means it has no revenue to pay for servers, monitoring, and upgrades. Some zero-commission validators are promotions that rise later, and some simply stop caring when the costs arrive. A modest, stable commission from an operator who has charged it for a year is usually a better sign than a zero rate that appeared last week. Also check whether the commission has changed recently: a validator that raises its rate sharply after attracting stake is telling you how it treats delegators.
Uptime and attestation history
Uptime is the record that matters most, and it is public. Every major chain has an explorer or staking dashboard that shows each validator's recent performance: blocks proposed, attestations made, and time offline. Look for months of steady history, not a good week. A validator that has been online for a year with rare, short gaps is a different proposition from one that launched last month with a perfect but meaningless record.
Pay attention to what the numbers mean on your chain. On some chains, missed attestations reduce your rewards quietly. On others, downtime can trigger penalties that reduce your stake. Our slashing guide explains the penalties and what triggers them. A validator with a clean slashing history and a public explanation for any past incident is worth more than one with no history at all, because every operator has a bad day eventually, and the honest ones tell you about it.
Self-stake: skin in the game
Self-stake is the amount of its own tokens the validator operator has bonded to its own node. It matters because it aligns incentives: an operator with meaningful self-stake loses its own money first if the validator is slashed or performs badly. A validator with almost no self-stake and a large pile of other people's stake has less to lose from its own mistakes. The right amount varies by chain, so compare validators against each other on the same chain rather than against an absolute number, and treat a large self-stake as a positive signal, not a guarantee.
Size cuts both ways. A very large validator is usually stable and well run, but delegating to the biggest validators concentrates the network, which is bad for the chain you are supporting. A mid-sized validator with good uptime and real self-stake is often the best balance: professional enough to stay online, small enough to care about each delegator. Spreading your stake across two or three validators instead of one also reduces the damage if any single operator has a bad month.
Reputation and communication
A validator is a service business, so judge it like one. Does it have a website or profile that explains who runs it and how to reach them? Does it post updates when it upgrades, when it has an incident, and when it changes its commission? Does it participate in the chain's governance or community, or is it a silent address with no public presence? An operator that communicates before an upgrade and explains itself after an incident is an operator you can trust with a problem, and problems are when trust matters.
Red flags are the mirror image: no way to contact the operator, no record of upgrades or incidents, a commission that changed without notice, or promises about returns that sound like a sales pitch. Remember that no validator can guarantee a return. Rewards depend on the chain, the total stake, and the validator's performance, and anyone promising a fixed yield is selling something else. Our staking rewards guide explains where rewards actually come from and what can reduce them.
Can you leave? Unbonding and switching
Before you delegate, check how you get out. Most chains have an unbonding period: a wait between asking to unstake and getting your tokens back, and it varies from hours to weeks depending on the chain. Some chains let you redelegate to a different validator instantly, others make you wait out the full unbonding period first. Read our unbonding period guide for the details, and confirm the current period on the chain's docs on the day, because it changes with network upgrades.
The practical rule is simple: never delegate tokens you might need next week, and never delegate your whole stake to one validator on your first try. Delegate a small amount, watch a full reward cycle, confirm the rewards match what the explorer predicted, and only then add more or move the rest. If the validator underperforms or stops communicating, redelegate. Switching costs a little time and sometimes a small fee, which is far cheaper than staying with a bad operator for a year.
The 5-minute checklist
- Commission: fair and stable for months, not zero and not recently raised without notice.
- Uptime: months of steady performance on the explorer, with rare and short gaps.
- Self-stake: meaningful own stake bonded, comparable to other validators on the same chain.
- Slashing history: clean record, or a past incident with a public explanation.
- Communication: reachable operator who posts upgrade and incident updates.
- Exit: you know the unbonding period and how to redelegate before you commit.
- Size: not one of the two or three largest validators if a solid mid-sized option exists.
A good validator is boring in the best way: same commission for months, same uptime, same updates, same operator answering questions. The checklist above finds those operators in five minutes, and it keeps working every time you delegate to a new chain. If you run a validator yourself and want your operation to be the one delegators pick from this checklist, see our node operations page, or start with a free infra audit and we will review your uptime, monitoring, and upgrade process against it.
Disclaimer: This post is educational only, not financial advice. Figures are listing or project data as of 2026-11-08 and may change - always confirm version numbers, requirements, and dates in the official docs before acting on them.
