Liquidation

Liquidation is the mechanism that keeps the protocol solvent when borrowers' collateral falls below a safe level. XPower Banq's liquidation differs from most protocols in one key way: the liquidator assumes the debt rather than repaying it.

This page covers liquidation from the borrower's perspective. For liquidators, see For keepers.

When liquidation triggers

Your position becomes liquidatable the moment your health factor drops below 100%. This can happen because:

• Collateral price falls.
• Debt-token price rises.
• Accrued borrow interest pushes your effective debt above the threshold.
• A combination of the above.

The protocol does not have a "warning" or "grace period." As soon as H < 100%, any keeper holding POOL_SQUARE_ROLE can liquidate via square(...), and — when governance has enabled the permissionless path by also granting POOL_SQUARE_ROLE to the pool itself — anyone can liquidate via liquidate(...) after paying the PoW cost.

What happens to you

A liquidation transfers a slice of your debt and a slice of your collateral atomically to the liquidator. The slice size is governed by the partial liquidation exponent e, which is passed in by the caller on every call — there is no protocol-stored default:

• e = 0 → 100% of your position (full liquidation)
• e = 1 → 50%
• e = 2 → 25%
• e = N → 1/2^N

Health factor under partial liquidation

For any e ≥ 1 (i.e. any partial liquidation), the slice s = 2^-e is taken uniformly from your debt and your collateral. Uniform slicing preserves the health factor exactly:

$$H_{\text{after}}=\frac{\alpha \cdot C\cdot (1-s)}{D\cdot (1-s)}=\frac{\alpha \cdot C}{D}=H_{\text{before}}$$

So a partial liquidation does not restore your H above 100%. If you were at H = 0.95 before a 50% slice, you are still at H = 0.95 after — your position is half the size, but no healthier. Another keeper can call liquidate(victim, ...) against you immediately; there is no protocol-imposed cooldown. Your H only rises if prices recover, interest swings, or you top up collateral / settle debt.

Health factor under full liquidation

e = 0 clears 100% of both sides at once. Both collateral and debt go to zero, so H = α · 0 / 0 is undefined (the position no longer exists in any meaningful accounting sense). From the protocol's point of view, the position is closed.

What you lose

At the liquidation boundary (H = 100%) with default weights, the implicit liquidation bonus is set by the weight ratio: (255/170) − 1 = 50%. A liquidator who takes debt of value $D$ at the boundary receives collateral worth $D\times 1.5$.

Strictly underwater (H < 100%), the realised bonus shrinks: uniform slicing always preserves the ratio of seized collateral to seized debt, and that ratio is your live C/D. The longer the keeper waits, the smaller their margin — at the extreme H ≤ α the liquidator breaks even or loses money, which is the bad-debt region.

In effect, at the boundary, liquidation costs you 50% of the debt being cleared, paid in collateral.

Worked example

You have 300 XPOW of collateral and 200 XPOW of debt. H = (300 × 0.667) / 200 = 100% — right at the edge.

A liquidator calls liquidate(victim, 1) (or square(caller, victim, 1) if they hold the role), clearing half:

• Debt cleared: 100 XPOW (50% of 200)
• Collateral transferred: 150 XPOW (50% of 300)

After:

• Your remaining collateral: 150 XPOW
• Your remaining debt: 100 XPOW
• New H = (150 × 0.667) / 100 = 100% — exactly preserved.

The liquidator's profit is 150 − 100 = 50 XPOW (the 50% boundary bonus, captured in collateral). Your position is half the size, but at the same H — and so it remains liquidatable for as long as your H stays under 100%.

Locked positions and liquidation

Liquidation works on locked positions too — locks don't make you immune. But there's a difference in what is transferred:

• For unlocked supply, the liquidator can immediately redeem the underlying tokens from the vault.
• For locked supply, the liquidator receives locked position tokens with the lock fraction preserved proportionally. They cannot redeem the underlying until the lock expires.

This is the mechanism behind cascade attenuation: only unlocked collateral can hit the open market, so cascade-driven sell pressure is bounded by the unlocked fraction.

In practice, liquidators prefer unlocked positions for the immediate liquidity. This creates de-facto liquidation seniority for locked positions: in a cascade, unlocked positions get hit first.

Two liquidation paths

The protocol has two ways to call liquidation:

1. square(user, victim, partial_exp) — the role-restricted path. Open only to holders of POOL_SQUARE_ROLE. No PoW gate.
2. liquidate(victim, partial_exp) — the public path. Anyone can call it, but it carries a per-partial_exp PoW cost from the POW_SQUARE parameter, and it works only when governance has granted POOL_SQUARE_ROLE to the pool itself. Internally it invokes this.square(msg.sender, victim, partial_exp), so the role check inside square() is applied to the pool — not to the external caller. Revoking the role from the pool address turns the public path off; granting it turns the public path on. Direct square() calls by other role-holders are unaffected either way.

Both paths use the same debt-assumption mechanics. See Debt assumption liquidation.

Avoiding liquidation

Three concrete strategies:

1. Maintain a safety margin. A higher H requires larger price moves to liquidate you. H ≥ 150% covers most normal volatility; H ≥ 200% covers most bear markets.
2. Top up collateral when H drops. Supply more, or settle some debt.
3. Set alerts. Off-chain monitoring tools can ping you when H falls below a threshold of your choosing.

The protocol does not auto-liquidate based on oracle staleness — there's a 12-hour TWAP half-life and a one-hour refresh interval, so the oracle responds slowly to genuine price moves. This works in your favour: a flash crash that recovers won't liquidate you, but a sustained crash will. See Oracle staleness.

Bad debt

If liquidation can't fully clear an underwater position — for example, in an extreme crash where the collateral value has fallen below the debt value — the protocol absorbs the loss as bad debt.

The over-collateralisation buffer (default 50%) is sized to absorb meaningful price moves before bad debt becomes possible. At the conservative governance floor (33% LTV with 200% buffer), simulation shows zero bad debt for crashes up to 50%. At the shipped 66.67% default, bad debt is bounded but non-zero in tail-event scenarios. See Bad debt scenarios.

Where to go next

• Debt-assumption liquidation — the mechanism in depth
• Health factor — the trigger condition
• Monitoring health — practical tips
• For keepers — running a liquidator