# Replacing the Banker with a Function ## A Mathematician's Guide to Decentralized Finance Shen-Ning Tung, National Tsing-Hua University <style>.reveal {font-size: 28px;}</style> --- ## I. The Global Picture — Architecture & Vision > Trust is not a leap of faith; it is the deterministic output of a distributed state machine. --- ### Defining the Transformation: Fintech vs. DeFi Before we dive into the math, we must distinguish between "digitalizing" a bank and "rebuilding" finance. * **Fintech (Digitized TradFi)**: Centralized institutions using modern tech stacks. You have an app, but the bank still holds the *"Master Ledger"* and the *"Kill Switch."* * **DeFi (Decentralized Finance)**: A peer-to-peer ecosystem where *Smart Contracts* act as the custodian and clearinghouse. * **The Paradigm Shift**: Moving from *Subjective Trust* (Law/Reputation) to *Objective Truth* (Cryptographic Proofs). ---  [What is DeFi (Decentralized Finance)?](https://whiteboardcrypto.com/defi-decentralized-finance/) [DeFi Beyond the Hype](https://wifpr.wharton.upenn.edu/wp-content/uploads/2021/05/DeFi-Beyond-the-Hype.pdf) [The Cambridge Centre for Alternative Finance](https://ccaf.io/defi/ecosystem-map/visualisation/graph) --- ### The Blockchain as a Global State Machine For a mathematician, a blockchain is simply a distributed, deterministic transition system. * **The State ($S$)**: A snapshot of all balances and contract data at time $t$. * **The Transition Function ($f$)**: The rules of the protocol (the code). * **The Formula**: $S_{t+1} = f(S_t, \Delta)$ * $\Delta$ represents a block of transactions. * Integrity is maintained because thousands of nodes must compute the same $S_{t+1}$ and reach consensus. ---  [What is a Blockchain?](https://whiteboardcrypto.com/what-is-a-blockchain/) [The tl;dr of Blockchains/Web3](https://youtube.com/playlist?list=PLEGCF-WLh2RKVT7xUwu0dCBGo3KH51XlS&si=6qBfs6aO3heUmU01) [Dan Boneh: Blockchain Primitives: Cryptography and Consensus](https://youtu.be/7Cu8KQmUhu0?si=YQnixg-Z6kg9Xd8i) --- ### The Engine of DeFi: Ethereum & The EVM <div style="font-size: 32px;"> To turn a ledger into a financial system, we need a Universal Computer capable of executing arbitrary logic. * **Smart Contracts**: Self-executing programs where the terms of the agreement are written directly into lines of code. They are autonomous (no middleman) and *immutable* (cannot be changed once deployed). * **The Ethereum Virtual Machine (EVM)**: A global, decentralized CPU. It ensures that every node in the network executes the same code and arrives at the same result—guaranteeing *mathematical determinism*. * **Gas & Halting Problem**: To prevent infinite loops in a Turing-complete system, every operation costs "Gas." This serves as a resource allocation mechanism and a formal constraint on computation. </div> ---  [Ethereum & Bitcoin Transaction Visualizer](https://txcity.io/v/eth-btc) [What is Ethereum?](https://ethereum.org/en/what-is-ethereum/) [What Are Smart Contracts and How Do They Work?](https://chain.link/education/smart-contracts) --- ### Oracles: The Bridge to Reality A closed mathematical system cannot natively "know" the price of Gold or the outcome of a game. * **The Oracle Problem**: How do we bring external data into a deterministic system without introducing a single point of failure? * **Decentralized Oracles**: Systems like *Chainlink* use a committee of nodes to provide a "consensus" data feed. * **Application**: This allows a smart contract to trigger a *liquidation* or a payout based on real-world events (e.g., $ETH$ price dropping below $1800$). ---  [What Is a Blockchain Oracle?](https://chain.link/education/blockchain-oracles) [What Is Chainlink? A Beginner’s Guide](https://blog.chain.link/what-is-chainlink/) --- ### Summary: The Architecture of Integrity | Feature | Traditional System | DeFi System | | :--- | :--- | :--- | | Enforcement | Legal Prose (Ambiguous) | Mathematical Code (Precise) | | Custodian | Centralized Bank | Self-Custody / Smart Contract | | Verification | Post-hoc Audits | Real-time Cryptographic Proofs | | Interoperability | Siloed APIs | Open Composability ("Money Legos") | <small>**The "Money Lego" Framework**: DeFi protocols are modular by design. Layer 1 provides settlement, tokens serve as raw materials, and lending/trading protocols act as building blocks. Developers "snap" these bricks together to create complex strategies without seeking institutional permission.</small> ---  [What is DeFi (Decentralized Finance)?](https://whiteboardcrypto.com/defi-decentralized-finance/) [DeFi Beyond the Hype](https://wifpr.wharton.upenn.edu/wp-content/uploads/2021/05/DeFi-Beyond-the-Hype.pdf) [The Cambridge Centre for Alternative Finance](https://ccaf.io/defi/ecosystem-map/visualisation/graph) --- ## II: Market Mechanics — How DeFi Works > Liquidity is an invariant, and price is the slope of a curve defined by the balance of reserves. --- ### Beyond Order Books: Automated Market Makers (AMMs) <div style="font-size: 28px;"> In traditional markets, a trade requires a counterparty (Buyer $\leftrightarrow$ Seller). In DeFi, you trade against a Liquidity Pool. * **The Liquidity Pool**: A smart contract holding a pair of tokens (e.g., $X$ and $Y$). * **The Mathematical Rule**: The pool uses a *Constant Product Invariant*: $$x \cdot y = k$$ * $x$: Quantity of Token A * $y$: Quantity of Token B * $k$: A constant that must be maintained during a trade. * **The "Peer-to-Pool" Model**: Anyone can become a *"Liquidity Provider" (LP)* by depositing assets, earning a pro-rata share of trading fees. </div> ---  [What Are Automated Market Makers (AMMs)?](https://chain.link/education-hub/what-is-an-automated-market-maker-amm) [Mastering AMMs: Guide to Automated Market Makers](https://threesigma.xyz/blog/defi-automated-market-maker-guide) [Mastering AMMs: Exploring Order Books and Intents](https://threesigma.xyz/blog/defi-automated-market-maker-guide) --- ### The Geometry of Price Discovery <div style="font-size: 28px;"> Price is not "set" by an admin; it is a mathematical consequence of the reserve ratio. * **The Marginal Price**: Defined as the negative slope of the curve at a specific point: $y = k/x$: $$P = \frac{dy}{dx} = -\frac{y}{x}$$ * **Slippage**: Because the curve is a hyperbola ($y = k/x$), every trade shifts the ratio $y/x$. * *Infinitesimal trades*: Price remains near the spot $y/x$. * *Large trades*: Significant displacement along the curve results in an exponential increase in the effective price. * **Asymptotic Liquidity**: The pool can never be fully depleted of an asset; as quantity approaches zero, the price approaches infinity. </div> --- ### Worked Example: Trading on the Curve * **Initial State**: Pool holds *100 ETH* ($x$) and *200,000 USDC* ($y$). * **The Invariant**: $k = 100 \times 200,000 = 20,000,000$. * **The Spot Price**: $P = y/x = 2,000$ per ETH. * **The Trade**: A trader wants to buy 10 ETH. * *New ETH reserve ($x'$)*: $100 - 10 = 90$ ETH. * *New USDC reserve ($y'$)*: $20,000,000 / 90 = 222,222.22$ USDC. * **The Cost**: Trader must pay $222,222.22 - 200,000 = 22,222.22$. * *Effective price*: \$2,222.22 per ETH. * *Slippage*: $\approx 11.1\%$ slippage from the initial spot price. ---  [Uniswap University](https://uniswap.university/) --- ### Algorithmic Lending & Money Markets Protocols like **Aave** replace loan officers with a *Utilization-based Interest Rate Model*. * **The Utilization Rate ($U$)**: Measures the capital efficiency of the pool.$$U = \frac{\text{Total Borrows}}{\text{Total Liquidity}}$$ * **Equilibrium Logic**: * *Low $U$*: Excess supply $\rightarrow$ Low interest rates to incentivize borrowing. * *High $U$*: Scarcity $\rightarrow$ Sharp interest spikes to attract lenders and force debt repayment. --- ### The Kinked Interest Rate Model The rate function $R(U)$ is a piecewise linear function centered around an *Optimal Utilization* ($U_{opt}$): * **The "Healthy" Zone ($U < U_{opt}$)**: Rates rise gently to manage normal demand.$$R_t = R_0 + \left( \frac{U_t}{U_{opt}} \right) R_{slope1}$$ * **The "Emergency" Zone ($U \geq U_{opt}$)**: Rates spike aggressively to protect protocol liquidity.$$R_t = R_0 + R_{slope1} + \left( \frac{U_t - U_{opt}}{1 - U_{opt}} \right) R_{slope2}$$ * **The "Kink"**: $R_{slope2}$ is usually an order of magnitude higher than $slope1$ (e.g., jumping from $4\%$ to $100\%$). This acts as a mathematical "emergency brake" to prevent pool depletion. ---  [DeFi Money Markets 2024](https://threesigma.xyz/blog/defi-money-markets-2024-guide) [Aave docs](https://aave.com/docs) [Compound docs](https://docs.compound.finance/) --- ### Risk Management: Over-collateralization In a pseudonymous system without legal recourse, all credit is secured by a *Collateral Invariant*. * **The Collateralization Ratio ($CR$)**: $$CR = \frac{\sum (\text{Collateral Assets} \times \text{Price})}{\text{Total Debt Value}}$$ * **Loan-to-Value (LTV)**: The maximum leverage allowed (e.g., $LTV = 75\%$ implies a minimum $CR$ of $133\%$). * **The Oracle Dependency**: The system's safety is strictly bound to the accuracy and latency of external price feeds. --- ### The Liquidation Function The "fail-safe" mechanism that maintains system-wide solvency. * **Trigger Condition**: If $CR < \text{Liquidation Threshold}$, the position is flagged as "under-collateralized." * **The Forced Exit**: Automated "Liquidator" bots repay the debt in exchange for the collateral at a Liquidation Bonus (e.g., a $5\%-10\%$ discount). * **Mathematical Solvency**: By liquidating before $CR < 100\%$, the protocol ensures that every dollar of debt is always backed by more than one dollar of market value. --- ### Liquidation Walkthrough: A "Hard Money" Event 1. **Open Position**: Deposit 1 ETH (\$2,000). Borrow 1,500 USDC ($LTV = 75\%$). 2. **Current State**: $CR = 2,000 / 1,500 = 133.3\%$ *(Healthy)*. 3. **Market Shock**: ETH price drops to *$1,700*. 4. **Breach**: $CR$ becomes $113.3\%$. If the threshold is $120\%$, the position is now *liquidatable*. 5. **Execution**: A bot detects the breach, repays the 1500 USDC, and claims the ETH at a discount. **The result**: Debt is cleared by code, not by credit committees or bailouts. The math is the ultimate arbiter of solvency. ---  [What Are Stablecoins and How Do They Work?](https://www.gemini.com/cryptopedia/what-are-stablecoins-how-do-they-work) [The Sky Protocol: Sky's Multi-Collateral Dai (MCD) System](https://makerdao.com/en/whitepaper/#introduction) --- ## III. Synthesis — Mechanism Design & The Future > The bedrock of future finance is not the institution, but the formally verified proof. --- ### Mechanism Design: Engineering Rational Behavior In a permissionless system, we cannot "ban" bad actors; we must make it mathematically unprofitable to be one. * **Agent-Based Economics**: We model the protocol as a game where autonomous agents maximize their own utility. * **Slashing & Incentives**: * *Positive Incentives*: Yield and rewards for "honest" participation (following the protocol). * *Negative Incentives*: Malicious behavior (e.g., double-spending or downtime) results in the automated forfeiture of "staked" capital (*Slashing*). * **Sybil Resistance**: Requiring a "Proof of Stake"—an economic cost that makes it prohibitively expensive for a single actor to subvert the system via multiple identities. --- [8 reasons why blockchain mechanism design is hard](https://a16zcrypto.com/posts/article/8-reasons-why-blockchain-mechanism-design-is-hard/) [Permissionless Mechanism Design](https://www.gauntlet.xyz/resources/permissionless-mechanism-design) [Blockchain Protocols and Web3: A Glimpse Under the Hood](https://youtu.be/-mWjDgVWl20) --- ### When the Math Breaks: The Terra/Luna Collapse <div style="font-size: 28px;"> Theory must meet reality. In May 2022, the Terra ecosystem provided a $40 billion case study in mechanism failure. * **The Setup**: TerraUSD (UST) was an algorithmic stablecoin. Unlike DAI (over-collateralized by ETH), UST was backed by a game-theoretic arbitrage loop with its sister token, LUNA. * **The Death Spiral**: A loss of confidence triggered a feedback loop: * UST drops below $1 $\rightarrow$ Users burn UST to mint LUNA $\rightarrow$ LUNA supply hyper-inflates $\rightarrow$ LUNA price crashes $\rightarrow$ UST loses more backing. * **The Lesson**: The system was *locally stable* but *globally fragile*. It assumed a constant presence of rational arbitrageurs and failed to account for "bank run" dynamics where liquidity vanishes instantly. </div> [Luna Brothers, Inc.](https://www.bitmex.com/blog/luna-brothers-inc) --- ### Privacy via Zero-Knowledge Proofs (ZKP) The "Transparency Paradox": Blockchains are public, but finance requires privacy. * **ZKP Definition**: A cryptographic method where a *Prover* convinces a *Verifier* that a statement is true (e.g., "I am solvent") without revealing the underlying data (e.g., "my balance"). * **Mathematical Applications**: * *Private Credit*: Prove your $Collateral > Debt$ without revealing your wallet's full history. * *Identity*: Prove you are a verified citizen or >18 without sharing your name or ID. * *Scalability (zk-Rollups)*: Bundling 10,000 transactions into a single "Succinct Proof" that the main chain can verify in milliseconds. --- [Zero Knowledge Proofs: An illustrated primer](https://blog.cryptographyengineering.com/2014/11/27/zero-knowledge-proofs-illustrated-primer/) [Zero Knowledge Proofs: An illustrated primer, Part 2](https://blog.cryptographyengineering.com/2017/01/21/zero-knowledge-proofs-an-illustrated-primer-part-2/) [Decentralized Identity: The Ultimate Guide 2025](https://www.dock.io/post/decentralized-identity) --- ### Open Problems: The Frontiers of DeFi This is an invitation for students to apply advanced mathematics to unsolved challenges: * *Formal Verification*: Using formal methods (e.g., TLA+, Coq) to prove a smart contract is bug-free before it handles billions of dollars. * *Systemic Risk & Contagion*: Modeling the "Money Lego" effect—how the failure of one protocol propagates through the hyper-connected liquidity network. * *MEV (Maximal Extractable Value)*: Studying the game theory of transaction ordering to prevent sophisticated bots from "taxing" users through front-running. --- ### Your Career: Beyond the Classroom Digital Finance is a rare field where pure math translates directly into production infrastructure. * **Protocol Architect**: Designing the next generation of invariant curves and interest rate models. * **Risk Quantitative (Quant)**: Running Monte Carlo simulations for firms like [*Gauntlet*](https://www.gauntlet.xyz/) or [*Chaos Labs*](https://chaoslabs.xyz/) to stress-test protocol parameters. * **The Opportunity**: You are at the intersection of Cryptography, Game Theory, and Distributed Systems. The "Wild West" era is ending; the *"Era of Rigor"* is beginning. --- ## Final Closing Thought > Digital Finance is not about 'disrupting' banks; it is about building a financial system that is as predictable and transparent as the laws of mathematics.