Staking
Proof of stake pays you for helping a blockchain agree with itself. Doing that through an exchange means handing the exchange the asset — and this one holds nothing at all. Here is the mechanism in full, and the specific reason it cannot ship here.
A yield is a wage, and somebody has to hold the key
A proof-of-stake chain picks who proposes and checks each block in proportion to how much value is bonded behind them, and pays those validators for the work. Collecting that payment takes two things: an asset bonded on the chain, and a key that signs with it. Every version of staking you can buy anywhere is an answer to one question — who holds those two things.
On this site the answer is nobody. No signing key exists here and nothing has ever been bonded; the only chain machinery in the system is a settlement rail confined to test networks, and a deposit here is a credit the system issues to itself. So there is no validator to bond to, no consensus reward to arrive, and no percentage that could go on this page without being made up.
On this page
Where a staking reward actually comes from
The chain itself pays it. Each block carries newly issued units and the fees the transactions in it were willing to pay, and both go to whoever secured that block. There is no borrower on the other side and no counterparty who owes anything — the units did not exist a moment earlier.
That is the part most rate tables leave out. Issuance comes out of everyone’s share: a holder who does not stake is diluted by the whole of it, and a staker’s real return is only the amount their nominal yield beats that dilution by. It is always a smaller and duller number than the one on the tile, and it is the one that describes what happened to your ownership.
A validator is a business, not a pipe. It keeps a commission before passing anything on, and it can be slow, offline or provably dishonest — each of which costs the stake behind it differently. An exchange sits one step further down again and takes its own cut of what is left, which is why the rate on an exchange staking page is never the rate the chain paid.
Three ways a yield can be paid, and which of them exist here
Every yield is somebody paying somebody. Naming who, for each of the three, is the fastest way to see why one of them is buildable here, one already runs, and one is not a matter of engineering effort.
A borrower pays it
Not builtSomeone posts collateral, draws your asset and pays interest for the use of it. The rate is an output of how much of the pool is borrowed. This one is missing a half that could be built, which is why the lending page shows no rate either.
LendingThe other side of the trade pays it
Live herePerpetual funding. On a real venue the long and short sides of a contract exchange a payment every eight hours, sized by a published rate. This product charges it for real at that same rate: an open position’s funding posts to the ledger at each crossing, and comes out of the position’s margin when the available balance cannot cover it.
Futures terminalThe chain issues it
Cannot exist hereStaking. New units minted by a protocol and handed to whoever secured the block. Nothing in this system is bonded to a protocol that could mint anything, and bonding it would mean custodying a real asset.
A staked balance has four states, and two of them pay nothing
Bond, earn, unbond, withdraw. Lending has the first and the last and nothing in between, which is the difference the drawing is for — the lower lane is the same money supplied to a lending pool over the same stretch of time.
Bonded, and not yet earning
The asset is committed and the chain has not started counting it. On Ethereum this is an entry queue that has run for weeks through 2026; on Cardano and Solana it is the wait for the next epoch boundary. Nothing about the price pauses meanwhile.
Unbonding, and this is the one that costs
You have asked for it back. It has stopped earning, it is still exposed to every move in the market, and you cannot sell it. A lender in the same position has already withdrawn. This gate is where the difference between the two products is actually paid, and it is where every staker caught by a drawdown was standing.
The exit is the product
Four networks this venue lists, on one scale. Cardano makes you wait for nothing; Ethereum makes you wait in a queue nobody controls. Calling all four “staking” hides the only parameter that will ever matter to you on a bad day.
| Network | When it starts earning | Getting back out | What can be taken from you |
|---|---|---|---|
| ADACardano | Delegation takes effect at the next epoch boundary, and an epoch is five days. The first reward arrives two or three epochs after that. | There is nothing to unbond. Delegated ADA never leaves the wallet holding it and stays spendable the entire time it is earning. | None. The protocol has no slashing condition to apply. |
| SOLSolana | Stake warms up at the next epoch boundary. An epoch is roughly two to three days. | Deactivation completes at the end of the following epoch, and no more than a quarter of the total active stake may leave in any single one. | Not yet. The chain records provable duplicate blocks on-chain; the penalties that would follow are still moving through governance. |
| DOTPolkadot | Bonded DOT starts backing validators in the next era. | About forty-eight hours, since the July 2026 vote. The week before that vote it was twenty-eight days. | Validators, yes. Nominators were removed from slashing entirely by the same vote. |
| ETHEthereum | After an entry queue. Through 2026 that queue has been measured in weeks, not hours. | A rate-limited exit queue, then a withdrawal sweep. It has cleared in hours and it has taken weeks, depending on who else is leaving. | Yes. The validator is forced out, loses a small initial penalty — one part in 4,096 of its effective balance — and takes a second penalty eighteen days later that scales with how much stake was slashed alongside it. |
Read against each network’s own documentation and governance record on 6 August 2026. These are protocol parameters set by other people’s votes, not measurements this system took.
Why that table carries a date
On 7 July 2026, getting DOT back took twenty-eight days and a nominator shared their validator’s slashing. On 9 July it took about two days and they did not. Nothing about the asset changed — two governance referenda changed it. Every exchange that had published a DOT staking specification was wrong the next morning, had no vote, and found out the same way everyone else did.
That is the part of a staking product nobody sells you: its terms are downstream of four foreign systems that can rewrite them without asking. A published specification is not a promise. It is a reading, and it needs the date it was taken.
Slashing is narrower than it sounds, and worse than it looks
A validator can lose money two ways and only one of them is slashing. The two get conflated constantly, which makes the common mild failure sound like the rare severe one and hides what actually makes the severe one severe.
Being offline is not slashing
On Ethereum a validator that misses its duties leaks penalties on roughly the scale of the rewards it failed to earn. It is symmetric, it is boring, and it stops the moment the machine comes back. Almost all of the “staking risk” a delegator will ever meet is this, and it is closer to earning nothing than to losing anything.
Slashing is for provable contradiction
Signing two conflicting things at the same height — a claim the chain can prove nobody makes by accident. The penalty is a forced exit plus a cut of the stake, and it exists because the entire security argument of proof of stake is that attacking the chain has to cost the attacker.
The part that matters is the correlation. Ethereum’s slashing penalty scales with how much other stake was slashed in the same window, so one operator misconfiguring one machine is cheap and a thousand validators running the same bug at the same moment is ruinous. That is the whole reason “spread across operators” is real advice rather than a slogan — and it is exactly the advice an exchange staking product cannot follow on your behalf while also being one operator.
Whether a delegator shares the operator’s slashing at all is a per-network decision, and it moves. Cardano has never had slashing. Solana records the evidence and does not yet apply the penalty. Polkadot took nominators out of it entirely in July 2026. Three different answers, one word.
Four products wearing one word
“Staking” names four arrangements that differ in exactly one thing, and it is not the yield. It is who is holding the asset and who is holding the key while it earns.
| Arrangement | Who holds the asset | Who signs with it | What you hold at the end |
|---|---|---|---|
| Solo stakingrun the validator yourself | You | You | The bonded stake itself, on-chain and in your name. |
| Delegated stakingnominate someone else’s validator | You | An operator you chose | Your own on-chain delegation. On some networks a delegator shares the operator’s slashing; on others they never did, or no longer do. |
| Liquid stakinga token that stands in for the position | A contract | The protocol’s operator set | A token tracking the bonded position, sellable without unbonding — at whatever discount the market puts on it when everyone leaves at once. |
| Exchange stakingthe product this page would have been | The exchange | The exchange | A row in the exchange’s database, and a claim against the exchange. |
The first three leave you on the chain. The fourth takes you off it. Bonded stake at an exchange is not your position with a custodian attached — it is the exchange’s position, and what you hold is a claim against the exchange. Every hard question about an exchange staking product falls out of that one row, and only the last row is a thing an exchange can sell at all.
That row exists for good reasons, and pretending otherwise would be its own kind of dishonesty. Solo staking on Ethereum starts at thirty-two ether and a machine that has to stay up; every arrangement below it trades a piece of control for not having to do that. The exchange sits at the far end of that trade, which is a real convenience and a real transfer of risk, and the only wrong thing to do about it is not to say so.
There is nothing here to bond
This exchange custodies nothing of value. Market data is read from a live venue; money is a double-entry ledger of our own. A deposit is a credit the system issues to itself — the transfer debits an internal account so the books still balance, and that account’s growing negative balance is the running total of everything ever issued. The one part of the system that touches a chain — a settlement rail for deposits and withdrawals — is confined by configuration to test networks, where tokens carry no monetary value, and it has never bonded anything to anyone.
For most missing features that is an obstacle to engineer around. For staking it is the boundary itself. Lending needs a borrower, and a borrower is code. Staking needs a real asset bonded on a real network under a key somebody is holding — and this system deliberately holds no signing key: even its test-network payout rail ships without one, and refuses to run until custody exists somewhere better than a configuration file. Until real custody exists, staking cannot.
Perpetual funding shows what the honest version looks like. This product does charge a real recurring payment every eight hours, and its other leg is that same internal account — the money is paper here as it is everywhere. What makes it defensible is that the rate is not ours: it is the venue’s published funding rate, streamed in and mirrored. A staking reward has no feed behind it. Nothing here has ever bonded a unit of ADA, so there is nothing to measure and nothing to mirror.
So there is no rate on this page. Not a blank field waiting to be filled, not an estimate, not a range, not a countdown to one.
- Credit
- your ADA · available+ reward
- Debit
- which account?− reward
The ledger refuses the first line without the second: every transfer is checked against zero in each asset before it is written, and an unbalanced one throws rather than posting. There are 7 accounts that could go on it, and every one of them already means something else.
external— That is a depositThe contra account every deposit already debits, so its growing negative balance is the running total of what this system has issued itself. Route a reward through it and the reward is a deposit wearing a different label.
fees— That is a rebateThe house’s own take from trading. Paying a yield out of it is giving back trading fees, which is a fine thing to do and is not staking. Calling it staking would misname where the money came from.
available— That is fundingAnother user’s spendable balance. Charging one account to pay another is exactly how perpetual funding works here — but no user on this exchange owes you a consensus reward, because no consensus happened.
locked— Already spoken forOrder reservations and position margin. Every unit in it is committed to an order or a position that expects it back.
settlement— That is a chain depositTokens this exchange holds on a chain, credited only when a deposit confirms. Debiting it for a reward would assert we received tokens on a chain that nobody sent us — and holding a token is a different event from a chain creating one.
customer_funds— That is everybody else’s moneyThe aggregate this exchange owes its users, and the account that makes client-fund segregation checkable. Paying a reward out of it pays you with the other users’ balances.
gas— That is a fee we paidWhat the house spends to broadcast a transaction. It records money leaving for a network, which is the opposite direction from a reward arriving out of one.
The account that belongs on that line is a chain’s own issuance, and there is no bucket for that. The nearest one is settlement, which means tokens this exchange holds on a chain — a different event, and one that only becomes true when a deposit confirms. Nothing here has ever held a key, signed a block or bonded a unit of anything.
What these four assets can do here today
All four trade on this venue against USDT, on the same live book every other pair uses. You can buy them, hold them, convert them and move them between wallets. You cannot bond them.
| Market | Network | Last (USDT) | 24h |
|---|---|---|---|
| ADAUSDT | Cardano | 0.20060 | +0.30% |
| SOLUSDT | Solana | 104.5400 | +1.28% |
| DOTUSDT | Polkadot | 0.84300 | +0.60% |
| ETHUSDT | Ethereum | 2,453.05 | +0.78% |
Read from the ingest worker’s snapshot on the last rebuild. This page refreshes about once a minute; the terminal streams the same prices tick by tick. System status.
Test funds can be credited directly in USDT, BTC, ETH and SOL, so an ETH or SOL balance is one form away. ADA and DOT you buy on the book like anything else — which is the honest version of “supported asset” on a site that issues its own money.
- Three wallets, none of them a staking wallet
- Spot, futures and funding, with transfers between them settling as real ledger movements. On a live exchange a wallet like the funding one is where balances park between products; here it is somewhere to deposit into and transfer from, and nothing else.
- Seventeen kinds of transfer, none of which bonds
- Deposit, withdrawal, order reserve and release, trade settlement, wallet transfer, convert, margin reserve and release, PnL settlement, funding fee, liquidation — and five settlement kinds for the test-network chain rail: chain deposit, withdrawal reserve, settle and release, and a gas cost nothing writes yet. There is no bond, unbond or reward anywhere among them.
- Seven buckets, and “bonded” is not one
- Available, locked, external and fees, joined by settlement, customer funds and gas when the chain rail landed. Locked holds order reservations and position margin, and both come back the moment the order is cancelled or the position closed. A bonded balance is precisely the one that does not.
If you are going to stake somewhere else, ask these five
This product cannot answer any of them, which is the point of the page. A product that can, should be able to answer all five without a support ticket.
Who runs the validator, and what commission do they take?
A product that will not name the operator is asking you to take the yield on trust. The commission is the difference between what the chain paid and what reaches you, and it is knowable in advance.
Is the rate a measurement or a projection?
“What this paid last period” and “what we expect it to pay” are different claims and only one of them has already happened. Chains change issuance; validators go offline; fee revenue moves with congestion.
Is it quoted before or after every cut?
The chain pays a validator, the validator keeps a commission, and the platform keeps a share of the remainder. A headline number can be true at any of those three points and mean something different at each.
Whose balance is it while it is bonded?
Held for you, or owed to you. That distinction is invisible on every ordinary day and is the only thing that matters on the day the platform is in trouble — which is also the day you will want to be leaving.
Is the exit the network’s, or the platform’s?
A platform can add a queue the chain does not have, batch withdrawals on its own schedule, or pause them. Compare the terms you were given against the network’s published parameters; they are public, and they will not always agree.
Not built yet
Staking would bond a balance to a validator on a real network, credit it the share of that network’s issuance the validator passes on, and hold it through the network’s own unbonding period before it could be sold again.
Waiting on
- Custody — a chain wallet, a signing key and a validator to bond to. No signing key exists anywhere in this system today, by deliberate design, which is the argument this page is built around
- A ledger vocabulary for a bonded balance: a bucket that is neither available nor locked, and bond, unbond and reward transfers to move value through it
- A per-network parameter set that survives its own governance — Polkadot rewrote two of the four facts in the table above in July 2026
- A reward source with a feed behind it, so a published rate is a measurement of something that happened rather than a number written down