What the definition does give you is something measurable before you deposit a cent. The loss depends on how far the two prices diverge, not on how long you stay. The arithmetic below uses one standard two-asset pool and clearly labelled illustrative numbers, so you can see where the loss is created and what it would take to offset it.

• The loss is relative, not absolute. No one invoices you for it. It is the difference between your position and the same assets held outside the pool.
• The size of the move sets the size of the loss. A 2x move in one asset produces about 5.7% of divergence in a standard 50/50 pool; a 4x move produces 20%.
• Time is not the variable. A position held for a day and one held for a year at the same end prices carry the same divergence loss. Only the fees differ.
• Fees are the offset, and they are not guaranteed. Sometimes cumulative fees exceed the loss, sometimes they do not. You can compute the break-even before depositing.

What creates the loss

A constant-product pool has to keep the product of its two balances constant, which forces it to trade against the market at every price change. When one asset rises, arbitrageurs buy it out of the pool and leave the other asset behind, so the position ends up holding less of the winner and more of the loser.

The pool sells what is rising

This is not a bug or a mispricing. It is what market making is: you are the counterparty to people who want to swap. The pool sells the appreciating asset because that is the only way to keep the invariant intact, and it does so continuously without asking you.

The consequence is structural. Any position holding two assets in a fixed proportion will underperform a straightforward holding of those same assets whenever their prices move apart. The pool does not lose money to a hack or a bad trade; it loses the difference between a rebalanced portfolio and a static one.

Why the word impermanent misleads

The loss is called impermanent because it reverses if the price ratio returns to where it started. That is true and almost useless. If one asset doubles and stays doubled, the loss is as permanent as anything in this market, and waiting does not undo it — it only gives fees more time to accumulate against a gap that is not closing.

The size of the gap depends on the pool’s shape, not on its fee tier. The table below is a rough map of where divergence comes from in common designs.

Impermanent loss, computed for a 2x move

A 50/50 pool holding 1 ETH and 2,000 USDC at an ETH price of $2,000 is worth $4,000. The alternative is holding exactly those 1 ETH and 2,000 USDC outside the pool. Every figure below is illustrative arithmetic, not a forecast of any asset.

Because the pool’s balances must multiply to the same constant, the composition slides along a fixed curve. At any price P, the position holds 1 divided by the square root of k in ETH and 2,000 multiplied by the square root of k in USDC, where k is P divided by $2,000.

Two things are worth staring at. First, a doubling and a halving produce the same percentage loss, 5.72%, because the formula depends on the ratio rather than the direction. Second, the loss is mild for small moves and then accelerates hard: about 1% at a 25% move, 5.72% at a 100% move, 20% at a 300% move.

The loss does not arrive as a fee. It arrives as a position that is quietly worth less than the one you decided not to open.

None of this requires the price to crash. A position can lose to holding in a rally just as easily as in a drawdown, which is the part most first-year users miss.

What fee income has to beat

Fees are the only thing standing between the divergence loss and a net loss, and they accumulate on a different schedule. The loss is set by the end price; fees are set by volume over time.

The break-even, in days

If the gap is 5.72% of the position and the pool pays an illustrative 0.05% of position value per day in fees, break-even is roughly 114 days of uninterrupted volume. If the price moves 4x instead, the gap is 20% and the same fee rate needs about 400 days, which assumes volume does not fall — a strong assumption in a market that has just moved violently.

That comparison is the argument in one line. A large price move creates a loss that normal-volume fees may never catch. A small move creates a loss that a busy pool clears in weeks. Neither outcome is guaranteed, and incentive tokens change the arithmetic without changing the mechanics.

Concentrated liquidity makes impermanent loss larger

Concentrating liquidity into a narrow price range multiplies the fee rate you earn and multiplies the divergence loss over the same price move. There is no free lunch in the range width. You are picking a point on a curve that trades fee density against divergence.

A narrow range behaves like a leveraged position. Because almost all the capital sits near the current price, a move of a few percent shifts the composition far faster than in a full-range pool. The upside is real: for the same capital you may earn several times the fees while price stays inside the range.

The failure mode is equally mechanical. Once price exits the range, the position is 100% in the weaker asset, stops earning fees entirely, and still carries the full divergence loss of the move. The range did not protect you; it concentrated the bet.

The post Impermanent Loss, With Actual Numbers appeared first on LadyChain.

Most first-year borrowers meet the liquidation health factor only after it has already dropped below 1, which is the worst possible time to learn how it works. This piece walks through the ratio, then a liquidation price calculation from collateral, loan and threshold using a clearly labelled illustrative position, so the mechanics are visible before anything is at stake.

• The threshold belongs to the market, not to you. It is set per asset and per pool, and it can change while your loan is open.
• The ratio falls for two reasons. Collateral gets cheaper, or debt gets bigger through accrued interest. Both push you toward the same line.
• Liquidation is usually partial and always penalised. A liquidator repays part of the debt and takes collateral worth more than they repaid.
• Only the oracle price triggers anything. A better price on a single venue does not protect a position the protocol values differently.

What the liquidation health factor measures

The ratio compares the borrowable value of your collateral against your debt, and the position is liquidatable when it falls below 1. It is a buffer measurement, not a score. A health factor of 1.30 means the collateral could lose roughly 23% of its value before the buffer is gone, which is a more useful way to read it than the number itself.

The numerator: collateral times threshold

The market assigns each collateral asset a liquidation threshold, expressed as a percentage of its value that counts toward solvency. A threshold of 80% means only $800 of every $1,000 of collateral is treated as backing your debt. The remaining 20% is the market’s cushion against the cost of auctioning collateral quickly.

This is the number people confuse with maximum loan-to-value. The maximum LTV caps what you may borrow when you open the position. The threshold decides when the position may be closed. The distance between them is the buffer, and it is the only reason a position does not die the moment it is opened.

The denominator: debt, including interest

Debt grows even when the collateral does not. Interest accrues per block or per second depending on the market, and it is added to the balance you owe. A position that is comfortable today can drift toward the line over months of doing nothing at all.

That drift is why a health factor is not a static property of the deposit. It is a function of time as well as price, and only one of those two variables is visible on a chart.

The parameters that feed the ratio are set per market, which is why the same collateral can behave differently in two places. The table below is a map of what each setting controls.

A liquidation price from collateral, loan and threshold

The liquidation price is the collateral price at which the numerator and denominator meet. Take an illustrative position: 10 ETH of collateral at $2,000, a market liquidation threshold of 80%, a maximum LTV of 75%, and a 12,000 USDC loan.

The loan is 60% of the collateral value, comfortably under the 75% cap, and the health factor is 20,000 multiplied by 0.80, divided by 12,000, which is 1.33. The liquidation price is the loan divided by the collateral quantity times the threshold, so 12,000 divided by 8, which is $1,500.

Now add twelve months of interest at an illustrative 5% on the debt, with the collateral price unchanged. The debt becomes 12,600 and the liquidation price becomes 12,600 divided by 8, or $1,575. The collateral never moved, and the position still became easier to liquidate.

This is the single most underrated part of the mechanic. Time alone moves the line toward you, and it does so quietly, without a candle on any chart.

What a liquidation actually does

Liquidation is normally partial, penalised, and open to anyone. A keeper repays a portion of your debt — often capped by a close factor set per market — and receives collateral worth slightly more than the repayment. The surplus is the incentive that keeps the mechanism running without a help desk.

Continue the example at an ETH price of $1,450, where the health factor is 0.967. An illustrative close factor of 50% lets a liquidator repay 6,000 USDC and take 6,480 of collateral at an 8% bonus, which is 4.469 ETH. You are left with 5.531 ETH worth $8,020 against 6,000 of debt, and a health factor of 1.07.

You did not lose everything. You lost 480 to the bonus, kept a smaller leveraged position, and the market is now further from your line than it was. That is the designed outcome. The undesigned outcome is a fast market where the price keeps falling while the transaction confirms, and the position arrives at the next block already under the line again.

Why the oracle price is the only price that matters

Every number above is computed from an oracle feed, not from the last trade on whichever venue you happen to watch. If the oracle says the collateral is worth $1,450, that is the value the contract uses, regardless of a better quote somewhere else.

Oracles create their own failure modes. A feed that updates too slowly can leave a position liquidatable at a price the market has already left behind. A feed that updates too fast, or one built from thin liquidity, can produce a wick that triggers liquidations nobody could have reacted to. Both directions of error cost borrowers money, and neither is visible in the interface’s health factor until it happens.

A liquidation price is not a prediction about the market. It is a statement about what one contract will do at one price, and the contract does not care whether you agree.

Repairing a low liquidation health factor

Repair means adding collateral or repaying debt, and both are harder in practice than they sound. The moment the ratio is low is usually the moment you are least able to move funds: gas is expensive, bridges are congested, and the stablecoin you need is on the wrong chain.

Why starting conservative is cheaper

A loan at 40% LTV has roughly twice the buffer of the same loan at 60%, and the cost of that safety is measured in forgone yield on the capital you did not borrow. That cost is fixed and known at the start. The cost of repairing a position under stress is neither fixed nor known.

Adding collateral later also tends to happen at the worst available price, because the reason you are adding is that the market has moved. Borrowing less at the start is the only version of this decision you get to make calmly.

The post Health Factor: What Liquidation Actually Looks Like appeared first on LadyChain.

The staking vs lending comparison matters most in a user’s first years, because the headline rate is the least informative figure in either product. This piece separates the mechanics from the marketing, using illustrative numbers where arithmetic helps, and leaves the ranking of the two to you.

• Staking pays issuance and fees; lending pays borrower demand. Both pay in the asset you deposited, which hides how different the sources are.
• Staking yield falls as more people stake. The issuance is fixed, so the per-token rate is diluted by participation.
• Lending yield falls as fewer people borrow. A quiet market can push the supply rate close to zero without any failure occurring.
• The exit is part of the yield. An unbonding queue and a utilisation ceiling are both ways of being unable to leave.

Where each yield actually comes from

Staking and lending produce income from entirely different economic events, and that difference explains almost every other difference between them.

Staking: issuance and fees

A proof-of-stake network pays validators newly issued tokens, plus a share of transaction fees, for producing blocks and attesting to others. Delegators receive a portion of that. The payment is denominated in the network’s own asset, created by the protocol, and its size is a governance parameter rather than a market price.

Because the reward is issued rather than earned from outside the system, staking is dilution-sensitive. The total payout is roughly fixed, so every additional token staked reduces the rate earned by everyone already staked. The reward does not respond to demand.

Lending: borrower demand

A lending market pays suppliers out of interest paid by borrowers. That interest is a price for liquidity, set by a curve that rises as more of the pool is borrowed. When nobody wants to borrow, nobody pays, and the supply rate collapses toward zero.

This is the opposite dynamic. Staking yield is set by supply of stake; lending yield is set by demand for debt. One falls when the crowd arrives, the other falls when the crowd leaves.

Staking vs lending: dilution against utilisation

Two variables drive the two yields, and neither is under your control. For staking it is the share of supply participating. For lending it is the share of the pool borrowed, usually called utilisation.

Dilution, in numbers

Suppose a network issues 5% of supply per year to stakers — an illustrative figure, not a claim about any chain. If 50% of supply is staked, the gross staking yield is 5% divided by 0.5, or 10%. If 80% is staked, the same issuance spread over a larger base gives 6.25%.

The more useful calculation adjusts for the supply growth itself. At 50% participation your balance grows 10% while total supply grows 5%, so your share of the network grows by about 4.76%. At 100% participation your balance grows 5% and supply grows 5%, so your share does not move at all. The nominal yield is still positive; the real claim on the network is flat.

Utilisation, in numbers

Lending rates move along a curve rather than being divided among participants. With an illustrative two-slope curve — 0.5% at zero utilisation, rising linearly to 6% at 80%, then steeply toward 40% at 100% — the supply rate is roughly the borrow rate multiplied by utilisation, before any reserve cut.

The last row is the trap. The highest supply rate in the table appears at the moment the pool is almost entirely lent out, which is also the moment you may be unable to withdraw, because only 5% of the deposits are sitting there as liquidity. High utilisation pays well right up until it pays nothing at all.

A staking rate is diluted by participation; a lending rate is priced by demand. Neither is a return you can bank on, and both are quoted as though the other variable does not exist.

A worked comparison on the same capital

Put $10,000 into each product and the difference in what determines the outcome becomes concrete. On the staking side, at 50% network participation the illustrative 10% rate pays $1,000 a year, and at 80% participation the same capital pays $625. The asset did not change; only the crowd did.

On the lending side, the same $10,000 earns $277.50 at 60% utilisation and $37.50 at 20%. That is a 7.4x swing in income with no change in the underlying asset, no change in the contract, and no failure of any kind. Utilisation simply moved.

Both are also paid in the asset you deposited, which means the yield is a claim on more of the same thing rather than a conversion into something else. If that asset loses value, a 10% yield does not offset a 30% decline — it just makes the decline slightly smaller in unit terms.

The exit is part of the yield

Getting out is where the two products diverge most sharply, and the exit terms belong in the comparison as much as the rate does. A staking position usually requires an unbonding period measured in epochs or days, during which the tokens are neither liquid nor earning under your control. A lending position can usually be withdrawn at any time — unless utilisation is at the ceiling, in which case it cannot be withdrawn at all.

Those are two different ways of being stuck. The staking queue is predictable: you know roughly how long, you simply cannot shorten it. The lending freeze is conditional: most days it costs you nothing, and on the day it matters it may last until borrowers repay.

Liquid staking tokens and instant unstake services change the picture by adding a market in between, usually at a discount. The discount is the price of skipping the queue, and it widens exactly when the queue is longest.

What the staking and lending headline rate hides

The staking vs lending headline rate hides the risk that produced it, and in both products that risk is a specific mechanism rather than a general mood. Staking exposes principal to slashing — penalties taken from the staked balance for validator misbehaviour, and in some designs for correlated outages. Lending exposes principal to contract failure, oracle failure, and bad debt that the pool has to absorb.

Neither list is shorter than the other. They are simply different, and which one matters more depends on what you can afford to lose and how closely you can watch it.

The post Staking vs Lending: Two Very Different Yields appeared first on LadyChain.