What You Can Do By The End

• Price a swap in a constant product pool
• Explain collateral and liquidation
• Compare stablecoin designs
• Explain oracles and their attacks
• Assess a protocol's risks

DeFi In One Idea

• Contracts hold tokens and apply the rules.
• Everything settles in a single transaction.
• No account, only a wallet and a signature.
• No support desk can reverse a mistake.

From Order Book To Pool

• An order book lists buy and sell offers.
• An AMM pool holds reserves of two tokens.
• A formula sets the price.
• Anyone can swap at any time.

The Constant Product Rule

• Reserves x and y must keep x times y.
• Adding token A forces token B out.
• Bigger trades get worse prices.

pool: 1000 A, 1000 B   (k = 1,000,000)
trader adds 100 A -> pool 1100 A
B must be 1,000,000 / 1100 = 909.09
trader receives 90.91 B (not 100)

Simulating A Swap In Python

• Fee is taken from the input amount.
• Output follows from the reserves.
• Run it with several trade sizes.

def quote(dx, x, y, fee=0.003):
    dx_eff = dx * (1 - fee)
    return y * dx_eff / (x + dx_eff)

x, y = 1000.0, 1000.0
for dx in (10, 100, 500):
    print(dx, round(quote(dx, x, y), 2))

Liquidity Providers

• Providers deposit both tokens into a pool.
• They receive shares and part of the fees.
• Impermanent loss appears when prices diverge.
• Fees may or may not cover it.

Lending Pools

• Depositors supply tokens and earn interest.
• Borrowers lock collateral first.
• Rates rise as more of the pool is borrowed.
• No one needs to know who you are.

Overcollateralization And Liquidation

• Borrow 100, lock 150 of another token.
• If collateral falls, the loan turns risky.
• Anyone may repay debt and take collateral at a discount.
• Fast crashes can still cause bad debt.

collateral value  150  ->  120
debt              100
ratio 1.50 -> 1.20 (below limit)
liquidator repays debt, gets collateral + bonus

Stablecoin Designs

• Fiat-backed: reserves held by an issuer.
• Crypto-collateralized: overcollateralized loans.
• Algorithmic: incentives with little collateral.
• Each design fails in a different way.

Oracles Bring Outside Data

• Contracts see only on-chain data.
• A price oracle reports market prices.
• Liquidations depend on those prices.
• Bad data means bad liquidations.

Oracle Manipulation

• One small pool is easy to move.
• Attacker pushes the price, then borrows against it.
• Safer: many sources, time averages, independent reporters.
• Add limits on sudden price jumps.

1. flash-borrow a large amount
2. swap to inflate price in a thin pool
3. borrow against the inflated price
4. repay the flash loan, keep the profit

Composability And Flash Loans

• Contracts are public and share token standards.
• One transaction can chain many protocols.
• A flash loan must be repaid in the same transaction.
• A flaw in one piece can spread.

Layers Of Risk

• Code: bugs and upgrade keys.
• Economics: designs that break under stress.
• Data: oracle failure.
• People: admin powers and governance.
• Markets: slippage, liquidation, front-running.

A Review Checklist

• Read the docs and find who holds admin keys.
• Check audits and open findings.
• Look for time delays on upgrades.
• Look at track record and value held.
• Start small, or start on a test network.

Recap

• AMMs price swaps from reserves.
• Lending relies on collateral and liquidation.
• Stablecoins need real backing.
• Oracles and composability add risk and power.

Next Week: Scaling

• Why Ethereum fees rise when demand grows.
• Rollups, optimistic and zero-knowledge.
• Bridges and the layer 2 landscape.

Questions And Exercise

• Run the AMM script with three trade sizes.
• Find a collateral factor in a lending doc.
• Write your five-line protocol review.