Blockchain Midterm 2 Combined

Defi Characteristics

Smart Contract Foundation - services executed by transparent auditable code rather than centralized institutions

Open Access - anyone can access through interest

Always Available - protocols operate continuously (unlike regular banks with banking hours, holidays, or geographic restrictions)

FTX

Collapsed in November 2022

$8 billion lost

opaque accounting and centralized control

users locked out

Defi Protocols

all transactions visible on-chain

users maintain full control over assets

zero downtime or fund freezes

cryptographic certainty instead of institutional trust

all transactions, collateral ratios, and protocol balances are publicly auditable in real-time

can only lend to anon using only on-chain verifiable assets

Centralized Exchanges

function as black boxes

users must trust management with custody of funds

Composability

all DeFi protocols can be combined to create sophiscated financial products

permissionless composability, so innovation does not require approval from any central authority

Main Protocols: stablecoins, AMMs, lending, perpetuals, governance

Example: acquire stablecoins, supply stablecoins to lending pool and receive interest-bearing tokens, use tokens as collateral to borrow and trade on Uniswap or GMX (tokens still earn yield)

Volatility Problem

native cryptos (ETH, BTC) experience dramatic price swings

unsuitable for predictable loan payments, merchant payment processing, denominating financial contracts (stating what currency to use in contract)

Stablecoin Solution

digital cash pegged to fiat currencies (typically USD) or crypto or algorithm

price anchor (stable reference for defi protocols)

safe haven (preserves value during crypto crashes)

programmable (smart contracts and blockchain capabilities)

enables other DeFi primitives

Fiat-Backed

backed by cash (typically USD)  or short term U.S. treasury securities (loans to government) held in traditional bank accounts

1:1 to fiat

subject to off-chain aduitors

examples: USDC (Circle), USDT

risk: requires trusting centralized entity with reserves, issuers can freeze addresses and block transactions, subject to government intervention and compliance requirements

if price below $1, arbitrageurs buy cheap tokens (non USDC), redeem for USDC for $1, pocket difference, demand increases → price rises

if price above $1, mint new USDC and sell them, supply increases → price drops

Crypto-Collaterized

over-collateralized with crypto assets

up to 150% ratio ($150 ETH backs $100 DAI)

if collateral value drops below threshold, automatically liquidate collateral

user returns stablecoin plus stability fee to reclaim collateral

all data on-chain

example: DAI (MakerDAO)

if above $1, mint more DAI by opening more vaults or swap USDC for DAI to increase supply

if below $1, repay loans (burn DAI) and retrieve collateral or swap DAI for USDC to reduce supply

pros: more transparent, more decentralized, and more censorship resisitant than fiat-backed

cons: more complex than fiat-backed, capital-inefficient, code vulnerabilities could drain collateral, heavy reliance on USDC for Peg Stability Module (PSM) so not that decentralized, liquidation can cascade

Algorithmic

incentive mechanism adjusts supply based on demand

mint/burn relationship with governance token (LUNA for UST)

example: UST (Terra) → failed

risk: death spiral if market confidence breaks (mass withdrawal)

if above $1, burn token to mint stablecoin

if below $1, burn stablecoin to mint token

Traditional Order Books

requires matching buyers to sellers

constant order placement and cancellation

high frequency updates

high gas costs

AMM Solution

liquidity always available

no need to match orders

algorithmic pricing

single transaction

passive liquidity provision

AMMs

pool holds reserves of two tokens

liquidity providers deposit both in proportion to current price

invariant ensures some formula of reserves stays constant (sum, product, etc.)

price and supply adjusts to maintain invariant

always available, no operator needed

example: Uniswap (general purpose), Balancer (multi-asset pools), Curve (stablecoin-optimized)

Over-Collateralized Lending

lock up collateral more than loan value

if collateral price drops, collateral ratio drops (less for collateral)

borrow to maintain ETH exposure while accessing liquidity, anonymous, tax efficient

global, permissionless access to liquidity without personal info or delays

automatic liquidation if collateral drops below threshold

less capital efficient than traditional loans

Lending in Action

lenders deposit liquidity into protocol and receive interest-bearing tokens

lock collateral and borrow against it (interest goes to lenders)

interest rates adjust based on utilization (high demand leads to higher rates which lead to more lenders)

borrowers repay principal plus interest to unlock collateral

Perps

no expiration (can hold positions indefinitely)

long positions (rise) and short positions (fall)

periodic payments between longs and short to keep price anchored to spot price

example platforms: dYdX (order book-based), GMX (pool-based leverage), Hyperliquid (high-performance L1=low latency)

Purpose of Perps

offset spot holdings with short positions

express market views with leverage

arbitrage funding rates and spot prices (exploit difference in price)

gain exposure without holding underlying assets

Governance via DAOs

create proposals for protocol changes (fee adjustments, parameter updates, treasury spending, or code upgrades)

governance token holders votes on-chain (voting power proportional to token holdings)

changes execute automatically when passed

maximal transparency

Maximal Extractable Value (MEV)

reorder, insert, or censor transactions to extract additional profit

example: user buys ETH, bot detects pending transaction, bot buys before transaction, bot sells post transaction, bot profits from price movement

DeFi Stack

stablecoins, AMMs, lending, perps, governance

each entry builds on primitives before it

Peg Stability Module (PSM)

DAI price stabilized through 1:1 swaps with USDC

DAI also backed by U.S. treasury bonds

45% backed by USDC, 30% backed by bonds, 25% backed by crypto

Hybrid Approaches for Stablecoin

backed by real-world assets and crypto collateral

integrated with fiat

community control through DAO governance

transparent with regular audits and proof-of-reserves

Constant-Sum Market Maker (CSMM)

linear invariant

x + y = k, delta x = delta y

1:1 exchange

vulnerable to arbitrage

unable to react to external market price changes (if market price shift to not 1 Y per 1 X, arbitrageurs can exploit)

one token can get completely drained

unsafe in practice (not used in real-life)

1mil USDC → 1mil DAI (1.0000 price, 0% price impact)

Constant-Product Market Maker (CPMM)

hyperbolic invariant

x y = k, delta y = (delta x * y)/(x + delta x)

trades cause slippage

larger trades causes more deviations between prices

when pool prices deviate from external markets, arbitrage trades profit until pool’s ratio aligns with global market

slippage prevents pool drainage

used for crypto since no assumption of fixed ratio

1mil USDC  → 980392 DAI (0.9804 price, 1.96% price drop)

Slippage

difference between average execution price and marginal price at start of trade

Curve’s StableSwap

A(x+y) + D = AD + (D³)/(4xy)

A(x+y) provides flat pricing near equilibrium

(D³)/(4xy) ensures infinite liquidity at extremes

A is tuning parameter to control hybrid behavior (critical amplification coefficient)

D is total liquidity

low slippage when prices near 1:1

better capital efficiency (large trades have minimal price impact)

CPMM at extremes prevent pool drainage

used for stablecoin exchanges to allow for simple swaps

1mil USDC → 999990 DAI (0.99999 price, 0.001% price drop)

Amplification Coefficient (A)

tuning lever in stableswap

higher A value → flatter around equilibrium (closer to CSMM)

Loan-to-Value Ratio (LTV)

maximum borrowing against collateral

example: 80% LTV on ETH allows $800 loan against $1000 collateral

lower LTV for volatile assets

set by protocol governance

leveraged trading and accessing liquidity without selling assets → avoiding taxable events

Utilization Rate (U)

U = total borrowed / total available liquidity

low U → interest rates fall to encourage borrowing

high U → interest rates rise to incentivize repayment and attract liquidity

optimal rate ~80%

Kinked Interest Rate Model

balance capital efficiency with liquidity risk

below optimal U → interest rates rise slowly to encourage borrowing and maximize capital efficiency (gentle slope)

above optimal U → interest rates surge sharply to prevent pool drainage and attract new liquidity (steep slope)

Health Factor (HF)

HF = (total collateral value * weight avg liquidation threshold)/(total borrow value)

HF > 1 → safe

HF <= 1 → under-collateralized, eligible for liquidation

affected by collateral asset value decreasing, borrowed asset value increasing, taking on additional debt

improve HF by supplying more collateral or repaying debt

example: initial $10000 in ETH with threshold 80% for $6000 in GH, HF=10000×0.8/6000=1.33, ETH drops to $7000, HF=7000×0.8/6000=0.933, liquidate

Liquidation Process

triggered by borrower’s health factor falling below 1

automated liquidator bots monitor blockchain and identify vulnerable position

liquidator calls liquidate, repaying borrower’s debt for them

liquidator receives portion of borrower’s collateral at discount (liquidation bonus), usually 5-10%

previous example: HF=0.933 ($7000 ETH collateral, $6000 GHO debt), with 5% liquidation bonus

liquidator spends $6000 GHO, receives $6300 ETH, profits $300

protocol recovers $6000 GHO and remains solvent

borrower’s debt cleared, receives $700 ETH collateral, lost $6300 to liquidation

Blockchain Oracle

feeds external data to smart contracts

needed to trigger liquidation

has multiple independent sources

resistant to short-term manipulation

de-centralized with no single point of failure

Flash Loan

borrow with zero collateral

loan must be borrowed and repaid with same atomic transaction

reverts if lending contract did not get repaid

used for arbitrage, liquidations, collateral swaps

“money lego” for developers and traders

Atomic Transaction

all or nothing

if every step succeeds, transaction occurs

if any step fails, transaction reverts

Oracle Manipulation Attack

borrowing massive capital through flash loan

use funds to execute swap on low-liquidity DEX to artificially inflate price of token X

gives inflated token X to as collateral to lending protocol using manipulated DEX as oracle, borrows maximum ETH with falsely valued collateral

repays original flash loan, keeps stolen ETH, price of token X crashes, protocol no longer solvent

Oracle Security Strategies

Decentralized Oracle Networks (DONs) - aggregate price data from multiple independent sources

Time-Weighted Average Price (TWAP) - calculate average price over time window (30 minutes), resistant to short-term price spikes

Multi-Source Validation - price agreement across all sources needed before accepting value, detects and rejects outliers

Aave Protocol

liquidity protocol

choose between variable and stable interest rates

pioneered flash loans

supports diverse range of assets as collateral, including volatile and niche tokens

why derivatives exist

mature markets need tools to manage and speculate on volatility

derivatives enable strategies like hedging and leveraged speculation

instead of trading assets directly, trade a contract that derives value from the asset (an echo of the original object)

hedging

protect holdings from price drops

leveraged speculation

amplify market exposure

spot trading

spot trading:

• simplest form of trading of the tokens themselves

• direct exchange for immediate ownership with no expiration/complex mechanics

traditional futures

traditional futures:

• contract to buy or sell at a set price on a future data

• like preordering a phone (price now, transaction later)

• fixed expiration forces price convergence

short

• contract to sell at future time with fixed price now

long

• contract to buy at future time with fixed price now

perpetual futures

• crypto innovation

• futures contract that never expires

• hold positions indefinitely with high leverage

• make/lose money based on how perp price changes in the future

• need to anchor price to reality if no expiration

short profit formula: position size * (entry perp price - exit perp price)

long profit formula: position size * (exit perp price - entry perp price)

spot vs futures vs perpetuals

spot:

• no expiration

• low/no leverage

• direct ownership of asset

• price anchor by market price

traditional futures:

• fixed date expiration

• has leverage

• only contract ownership

• price anchor by expiration date

perp futures:

• no expiration

• high leverage

• contract only ownership

• price anchor by funding rate

why do perps need the funding rate mechanism

without expiration date, perp contract price can drift far from actual spot price making it meaningless

funding rate mechanism

periodic payment between long and short position holders

creates incentives that anchor perp prices to spot prices

1. when perp price > spot price

   • market is bullish with more longs

   • longs pay shorts, making long positions expensive and shorts profitable

   • reward shorts > pushes perp price down toward spot

   • funding positive

2. when perp price < spot price

   • market is bearish with more shorts

   • shorts pay longs

   • reward longs > push perp price up toward spot

   • funding negative

maintain contract relevance and utility by preventing perp price from drifting

funding rate formula

funding payments calculated every 8 hours

amount = position size * funding rate

funding rate = fixed interest rate + premium

premium = percentage difference between perp and spot prices (Pperp-Pspot)/Pspot averaged over time to prevent manipulation

interest component balances borrowing cost differences, reflect how usd changes?

positive funding rate example

setup:

• position: long 1 BTC

• spot price: $100,000

• perp price: $100,120 (bullish premium)

• position notional (entry size): $100,120

• interest: 0.01%

calculate premium:

• P = (100,120-100,000)/100,000=0.12%

calculate funding rate

• F = 0.12%+0.01%=0.13%

calculate payment

• payment = 100,120 × 0.13% = $130.16

double edged sword of leverage and liquidation

leverage

• main attraction of perpetuals

• control large positions with small capital deposits called margin

liquidation

• deposit automatically liquidated if loss is about to exceed the deposit

• if short, you are forced to buy

• if long, you are forced to sell

liquidation cascade

liquidation is a risk when many traders cluster at similar price points

after breaking through a liquidation wall, a cluster of traders are liquidated simultaneously and the exchange dumps all their positions causing the price to change drastically

new price can break through another liquidation wall and make another cluster of traders liquidate, causing a chain reaction

case study decentralized perps: GMX

• trader vs pool model

• pioneer on-chain perpetuals on layer 2 networks like arbitrum???

• pooled liquidity model centered around GLP token eliminating need for traditional order books

GLP

• multi-asset liquidity pool where users deposit variety of tokens into one pool and receive GLP tokens in return representing their share of the entire pool

trading mechanism:

• traders trade directly against GLP pool

• liquidity providers collectively act as single counterparty “house”

oracle base pricing

• GMX uses high speed oracles like chainlink that aggregate prices from major centralized exchanges

key feature: zero slippage, large orders do not impact price. use oracle price

profits and losses go to and from the GLP pool

liquidity providers betting that traders will lose more than they win, earning fees and trader losses as yield

case study decentralized perps: Hyperliquid

• new gen, central limit order book (CLOB) on own application-specific blockchain

• makes traditional exchange mechanics fully on-chain

CLOB:

• traditional model used by centralized exchanges/stock markets

• public ledge of all buy and sell orders at different price levels visible to all participants

trader-to-trader

• traders match against each other directly

• buy orders need to find corresponding sell orders

• market makers provide liquidity

native price discovery

• prices are discovered on-platform thorugh order matching

• market price is last matched trade where highest bid meets lowest ask

gmx vs hyperliquid trade off

gmx:

strengths

• zero slippage

• simple lp participation

• capital efficient

• predictable execution prices

weaknesses

• no native price discovery

• oracle dependency risk

• lps bear all trader P&L risk

• prices imported not discovered

hyperliquid:

strengths

• native price discovery

• traditional market dynamics

• no oracle dependency

• real-time order matching

weaknesses

• slippage on large orders

• requires market maker presence

• thin books > high slippage

• complex infrastructure needs

gmx vs hyperliquid comparison summary

gmx:

• liquidity from glp pool (collective lps)

• price from external oracles

• zero slippage because use oracle price

• counterparty: pool vs trader

• infrastructure uses L2 smart contracts

• best for predictable execution

hyperliquid:

• liquidity from individual market makers

• price from on-chain order matching

• variable slippage depending on order book depth

• counterparty: trader vs trader

• custom blockchain infra

• beset for price discovery and transparency

blockchain mempool

when user sends a transaction, it enters the mempool (memory pool)

• public waiting area for unconfirmed transactions

• each node maintains own mempool version, converging on similar pending transaction sets

• transparent: all transactions, contents, and gas fees are publicly visible

what is MEV

Originally “miner extractable value” in proof-of-work systems

evolved now “Maximal Extractable Value”

core definition:

• maximum value extractable by block producers through strategic inclusion, exclusion, or reordering of transactions within blocks they create, beyond standard block rewards and transaction fees

information asymmetry

• MEV is possible because transactions are public and can be reordered before confirmation

• block producers and “searchers” exploit this by monitoring the mempool for profitable opportunities

core MEV strategies

front-running

• executing transactions before victims by paying higher gas fees

back-running

• capitalizing on price movements immediately after target transactions

sandwich attacks

• combining front-running and back-running to trap victim transactions

front-running mechanics

• front-runner spots profitable pending transaction in mempool (like large DEX swap that will move an asset’s price)

• attacker submits own transaction with higher gas fee to get priority inclusion

• front-runner executes first, causing price to move up

• original transaction gets worse price, and attacker gets more value

back-running mechanics

• execute transaction immediately after a target transaction to profit from price impact

• ex: if large trade causes price imbalance between two DEXs, a back-runner can buy on cheaper DEX and sell on pricier DEX

• less harmful than front-running, doesn’t worsen victim’s execution, just profits off of it

sandwich attack

• first frontrun, then victim trades, then backrun

• one of most common and harmful MEV strategies

liquidation MEV: necessary evil

• in decentralized lending, when borrower’s health factor drops below threshold, their position is eligible for liquidation

• liquidators repay debt in exchange for the collateral at a discount (liquidation bonus)

• highly competitive process, with multiple liquidator bots engaging in gas bidding wars to try an get their transaction first (others fail)

• liquidation MEV is essential for protocol health

• ensures under-collateralized loans close quickly, protecting protocols from accumulating bad debt

oracle latency MEV

blockchain oracles vitally bring off-chain data to smart contracts, but are slow

oracles update every n minutes or after significant price deviations

latency problem

• delay between real-world price changes and on-chain oracle updates creates oracle latency

• window where smart contracts operate on outdated info

exploitation opportunity

• attakers monitor real-world prices and on-chain prices, exploiting discrepancies

• can have catastrophic consequences for protocols

  • ex: LUNA/UST 2022 collapse

case study: LUNA De-peg

during LUNA price crash, real-world prices on centralized exchanges like binance fell faster than on-chain oracles updated

attackers could buy buy a lot of tokens for real price and deposit into protocol where it is worth more under outdated oracle

protocol allows to borrow against that inflated collateral

after oracle updated, protocol got a lot of debt

pros/cons of MEV

pros

• price efficiency: arbitrage MEV maintains consistent pricing across different DEXs

• protocol solvency: liquidation MEV provides powerful incentives ensuring lending protocols remain solvent by quickly closing under-collateralized positions

• network incentives: MEV revenue can constitute a significant portion of block producer income, incentivizing rubust network security and participation

cons:

• frontrunning and sandwich attacks harm victims through slippage, execution prices, and losses

• network congestion: fierce competition among MEV bots create gas wars that inflate transaction fees and clog the network for everyone

• centralization risk: high-value MEV opportunities favor entities with more resources

MEV mitigation strategies

chain-level and protocol solutions

• modify fundamental blockchain rules or transaction supply chain to reduce harmful MEV on protocol level

application-level

• strategies that decentralized apps, especially DEXs use to protect their users from bad MEV attacks

private transaction relays

most widely adopted MEV mitigation

how works

• users send transactions directly to a private relay instead of public mempool

• relay forwards transactions to specialized block builders with agreements to include them

key benefit

• transactions are hidden from public MEV bots until included in a block

limitation

• users must trust relay and builder to not exploit themselves

• but reputation and economic incentives of major builders prevent bad behavior

centralization problem which proposer-builder separation (PBS) combats

• validators who propose blocks also build them

• to maximize profit, proposers run resource-intensive MEV extraction software, which favor large staking pools

• small solo-stakers cannot compete and profit less from their blocks since they make less MEV

• pressures users to stake with large pools, driving validator centralization

proposer-builder separation (PBS) solution

separate roles into two specialized actors

builders (specialists):

• highly specialized entities that build maximally profitable blocks with MEV before submitting bids to proposers

proposers (validators):

• select most profitable block from competing builders based on bid amount

• no need for MEV expertise

does not stop MEV, but acknowledges extracting MEV is difficult

builders must bid their potential profit in order to get their block built

money is moved from few specialists who find MEV to all validators who secure the network

proposer-builder separation (PBS) benefits

• solo stakers are as profitable as large pools since neither build the actual blocks, both accept the winning bids from open market

• no more pressure for validator centralization

• creates competitive building market that maximizes value returned to all stakers

slippage tolerance: application level MEV protection

• most common user-facing MEV protection

• when users make swaps, DEX interfaces require them to set a slippage tolerance

• if final price exceeds tolerance, the transaction is failed

• defends against sandwich profitability since they can only extract so much that it doesn’t fail the user’s transaction

• limitation since bots can still extract MEV without triggering limits with an upper bound

batch auctions: application level MEV protection

• some DEXs use this instead of sequential execution

• collection phase: protocol collects all user trades from short time window into a batch

• solving phase: third party solver examines trades in batch and calculates single uniform clearing price

• execution phase: everyone in batch receives same prices regardless of submission order

traditional AMMs are sequential where order matters, which attackers can exploit

defi stack

value (stablecoin) > exchange (AMM) > credit (lending) > derivatives (perps) > governance (DAOs)

what is a DAO

rules as code: organizational bylaws encoded in transparent, immutable smart contracts

member control: token holders collectively govern through binding, on-chain votes

no central authority: flat, democatic structure replacing traditional hierarchies

token-weighted voting

most prevalent DAO governance mechanism

voting power scales linearly with token ownership

allows voting power with financial stake

simple implementation, default choice for most DAOs

whale problem with token-weighted voting

when token distribution is unequal, a handful of large holders (whales) can dominate decisions, marginalizing thousands of smaller participants

plutocracy

quadratic voting

solution to wealth concentration by making additional votes exponentially expensive

cost to cast n votes equals n² credits

forces strategic allocation of credits when you vote

ex: if you have 16 credits, you can cast 1 vote on 16 different proposals or cast 4 votes on one critical issue

governance failure: flash loan attack

abuse flash loans to temporarily hijack voting power within single atomic transaction

1. borrow: attacker gets massive flash loan of governance tokens

2. vote: temporary voting power passes malicious proposal

3. execute: proposal executes immediately, draining treasury

4. repay: flash loan repaid, attacker walks away with funds

ex: Beanstalk Farms, because proposals were executed instantly after passing

solution to flash loan attacks

mandatory time delays between vote passage and execution to prevent instantaneous manipulation

curve wars

2021-2022 multi-billion dollar fight in DeFi to control Curve Finance

Curve: largest exchange for stablecoins and key part of how digital financial systems work

Big players like Convex Finance and Frax Finance constantly bought Curve’s main CRV token to get more voting power

Want to send CRV rewards to their own investment pools, attracting users and giving financial sway

Large amounts of money called Total Value Locked (TVL) at stake

example of how DAO governance works, show game theory in decentralized autonomous orgs

curve governance mechanism

CRV governance token can be locked up for up to 4 years to receive veCRV (vote-escrowed CRV)

veCRV holders voe on which liquidity pools receive weekly emissions of new CRV tokens as rewards

directing CRV rewards to specific pool attracts more liquidity as LPs chase maximum yield

deep liquidity is crucial for stablecoin protocols

control votes > control liquidity > control market

curve battle cycle

loop of increasing dominance

acquire CRV tokens > lock CRV for veCRV voting power > vote to direct CRV rewards to own pool > high rewards attract more liquidity providers to that pool > deep liquidity generates revenue > acquire more CRV tokens with extra fees