Why must you post more collateral than you borrow?

Every bank loan you have met rests on an assessment of the borrower: income, history, security, and — underneath all of it — a legal system that can pursue them. A DeFi lending protocol has none of that. It does not know who you are, cannot check anything, and cannot sue anyone. Strip all of it away and exactly one enforcement mechanism remains.

The collateral is the entire credit process

A permissionless lending market cannot assess creditworthiness, so it does not try. Instead it demands collateral worth more than the loan, and it holds that collateral inside the contract. If the borrower never repays, the protocol does not care — it sells the collateral. The loan is not underwritten; it is secured to the point where underwriting is unnecessary.

That is why the cushion must exceed 100%, and the size of the cushion is driven by four things:

1. Volatility. Collateral can fall between the moment a position becomes unsafe and the moment it is sold.
2. Liquidation latency. Selling requires a transaction, which requires a block, which costs gas. In congestion — see Black Thursday in the previous lesson — that delay lengthens exactly when it is most costly.
3. Liquidity of the collateral. A thin token cannot be sold in size at the screen price.
4. Oracle quality. The protocol acts on a reported price, not a real one (Unit 4).

The parameters, and the arithmetic

Two numbers govern a position, and mixing them up is the classic beginner error:

• Collateral factor (max LTV) — how much you may borrow at the moment of borrowing.
• Liquidation threshold — the level at which the position becomes liquidatable. Always ≥ the collateral factor; the gap between them is deliberate breathing room.

Worked. Deposit $10,000 of ETH. Collateral factor 80%, liquidation threshold 82.5%.

Maximum initial borrow = 10,000 × 0.80 = $8,000

Borrow a conservative $5,000 of a stablecoin instead. The liquidation price is where collateral × threshold equals debt:

collateral at liquidation = 5,000 ÷ 0.825 = $6,060.61

So the collateral can fall 39.4% before the position is at risk:

(10,000 − 6,060.61) ÷ 10,000 = 39.4%

Now borrow the full $8,000 instead:

collateral at liquidation = 8,000 ÷ 0.825 = $9,696.97 → a fall of just 3.03%

Same collateral, same asset, same protocol. The distance to liquidation went from 39.4% to 3.03% purely because of how much was drawn. The borrower sets the risk, not the protocol — and borrowing the maximum permitted is, arithmetically, borrowing at the edge of a cliff.

The second hand on the clock: interest

Debt is not static. Borrow interest accrues continuously and is added to the balance, so the safety margin erodes even if the price never moves. At a 6% borrow rate, an $8,000 debt grows to about $8,495 in a year — and the collateral needed to support it grows with it. A position parked and forgotten is not a position frozen.

Why anyone borrows against collateral they already own

Stated descriptively, not as suggestions: to obtain liquidity without selling (and, in some jurisdictions, without a disposal event — a tax question for a qualified professional, not for this course); to short an asset by borrowing and selling it; to lever exposure through the recursive loop from Unit 1; and to fund some other position. Every one of these converts a price risk into a price risk plus a liquidation risk plus a smart-contract risk, which is the honest description of what over-collateralised borrowing does to a balance sheet.

Try it now

1. Read the current price off the chart below and build a position on paper: $25,000 of collateral, 80% collateral factor, 82.5% liquidation threshold. Compute the liquidation price at 25%, 50% and 100% of the maximum borrow.

2. Now put your three numbers on the chart. Drop a Level at each liquidation price and look at how much of the last five years sits below them. The most aggressive of the three is not a hypothetical: the price has been there.
3. Measure the largest single-day fall in the window — the tallest red candle — and ask which of your three positions would have survived it. Then find one lending protocol's public parameter table, note how collateral factors differ across assets, and write one sentence explaining what the protocol is saying about each asset by choosing that number.