Key Takeaways
1. The Time Value of Money and No-Arbitrage form the bedrock of quantitative finance.
The simplest concept in finance is that of the time value of money ; $1 today is worth more than $1 in a year’s time.
Time value of money. Capital has a temporal dimension because it can be invested to earn interest over time. Continuous compounding, represented mathematically by $e^{rt}$, is the standard assumption in quantitative finance, allowing us to discount future cash flows back to their present value.
The no-arbitrage principle. This concept asserts that there are no "free lunches" in efficient financial markets. If two risk-free investment strategies yield different returns, market participants will instantly exploit the mispricing, driving the prices back into equilibrium.
Key applications of no-arbitrage:
- Determining forward and futures prices based on spot prices and interest rates.
- Establishing put-call parity, a model-independent relationship between calls and puts.
- Replicating derivative payoffs using portfolios of underlying assets and cash.
2. Asset prices are fundamentally random walks driven by volatility.
Almost all of sophisticated finance theory assumes that prices are random, the question is how to model that randomness.
Modeling market randomness. Stock prices do not move in predictable paths; instead, they exhibit random walks. The widely accepted model for equities, currencies, and commodities is the lognormal random walk, represented by the stochastic differential equation $dS = \mu S dt + \sigma S dX$.
The role of volatility. Volatility ($\sigma$) measures the standard deviation of asset returns and represents the intensity of market noise. While the drift ($\mu$) dominates over long horizons, volatility dominates short-term price movements and is the single most critical parameter in option pricing.
Stochastic calculus foundations:
- Itô's Lemma: The stochastic equivalent of Taylor series, used to differentiate functions of random variables.
- Wiener Process ($dX$): A continuous-time random walk with a mean of zero and variance of $dt$.
- Markov Property: The future state of the asset depends only on its current price, not its historical path.
3. Delta hedging dynamically eliminates risk to derive the Black-Scholes equation.
The Black–Scholes equation contains all the obvious variables and parameters such as the underlying, time, and volatility, but there is no mention of the drift rate µ.
Riskless portfolio replication. By combining a long position in an option with a short position in a specific quantity of the underlying asset (the delta, $\Delta = \partial V/\partial S$), an investor can eliminate all random price fluctuations. This process is known as delta hedging.
Deriving the Black-Scholes equation. Because the delta-hedged portfolio is completely risk-free over an infinitesimal time step, its return must equal the risk-free interest rate. This no-arbitrage condition yields the famous Black-Scholes partial differential equation: $\frac{\partial V}{\partial t} + \frac{1}{2}\sigma^2 S^2 \frac{\partial^2 V}{\partial S^2} + rS\frac{\partial V}{\partial S} - rV = 0$.
Risk-neutral valuation:
- The drift rate $\mu$ drops out of the pricing equation entirely.
- Options are priced as if the underlying asset grows at the risk-free rate $r$.
- The option value is the discounted expected payoff under this risk-neutral probability measure.
4. The Greeks are essential tools for managing risk and exposure.
It can be argued that it is more important to get the hedging correct than to be precise in the pricing of a contract.
Measuring portfolio sensitivities. The "Greeks" are the partial derivatives of the option value with respect to its underlying variables and parameters. They allow traders to quantify exactly how their portfolios will react to changes in the market environment.
Key risk metrics. Delta ($\Delta$) measures sensitivity to the asset price, while Gamma ($\Gamma$) measures the sensitivity of Delta itself, indicating how frequently a portfolio must be rehedged. Theta ($\Theta$) represents time decay, and Vega measures sensitivity to changes in volatility.
Hedging applications:
- Delta Neutrality: Maintaining a portfolio where the net Delta is zero to eliminate directional risk.
- Gamma Neutrality: Using other options to make the net Gamma zero, reducing the frequency of rehedging.
- Vega Hedging: Offsetting volatility risk by trading options with offsetting Vega exposures.
5. Early exercise transforms American options into complex free-boundary problems.
The American option value is maximized by an exercise strategy that makes the option value and option delta continuous
The optionality of early exercise. Unlike European options, American options can be exercised at any time prior to expiry. This added right means American options can never be worth less than their European counterparts, and they introduce the challenge of determining the optimal exercise boundary.
Free-boundary mathematical formulation. The boundary separating the continuation region (where the option is held) from the exercise region (where it is exercised) is not known in advance. It must be solved as part of the system, turning the linear Black-Scholes equation into a non-linear obstacle problem.
Key characteristics:
- Smooth Pasting: The option value and its delta must be continuous across the early exercise boundary.
- Dividend Impact: It is never optimal to exercise an American call early on a non-dividend-paying stock.
- Numerical Necessity: American options generally lack closed-form solutions and must be solved using finite-difference or binomial methods.
6. Volatility is not constant; it smiles, skews, and behaves stochastically.
The implied volatility is the volatility of the underlying which when substituted into the Black–Scholes formula gives a theoretical price equal to the market price.
The volatility smile. If the Black-Scholes model were perfect, all options on the same asset would have the same implied volatility. In reality, plotting implied volatility against strike price reveals a "smile" or "skew," showing that the market prices in the probability of extreme events.
Local and stochastic volatility. To capture the smile, quants use local volatility models, where volatility is a deterministic function of asset price and time, $\sigma(S,t)$. Alternatively, stochastic volatility models treat volatility as a separate random variable governed by its own SDE.
Modeling approaches:
- Local Volatility Surface: Calibrating $\sigma(S,t)$ to perfectly match all traded vanilla option prices.
- Stochastic Volatility (e.g., Heston): Introducing a second random factor, requiring a market price of volatility risk.
- Uncertain Volatility: Assuming volatility lies within a band $[\sigma^-, \sigma^+]$ and pricing via worst-case scenarios.
7. Real-world frictions like transaction costs and discrete hedging break the Black-Scholes ideal.
Because the total hedging error does disappear as the time between rehedges shrinks we can easily argue for the validity of the Black-Scholes equation, but only in that limit.
The reality of discrete hedging. Continuous rehedging is physically impossible and financially ruinous. Hedging at discrete intervals introduces a random "hedging error" that is proportional to the option's Gamma and the time step, meaning risk can never be completely eliminated.
The impact of transaction costs. Buying and selling the underlying asset incurs transaction costs due to bid-ask spreads. In the presence of these costs, the Black-Scholes replication strategy leads to infinite costs in the continuous limit, requiring a modified, non-linear pricing equation.
Key non-linear models:
- Leland's Model: Adjusting the volatility parameter based on the transaction cost rate and the hedging frequency.
- Hoggard-Whalley-Wilmott: Extending Leland's model to portfolios, where the portfolio value is no longer the sum of its parts.
- Utility-Based Hedging: Finding a hedging bandwidth that maximizes the investor's expected utility under transaction costs.
8. One-factor interest rate models must balance tractability with yield curve calibration.
The 'spot rate' that we will be modeling is a very loosely-defined quantity, meant to represent the yield on a bond of infinitesimal maturity.
Modeling the short rate. In fixed-income modeling, we cannot easily hedge with the underlying because the short-term interest rate ($r$) is not a traded asset. This requires the introduction of the "market price of interest rate risk" ($\lambda$) to derive the bond pricing equation.
Tractable short-rate models. Models like Vasicek and Cox-Ingersoll-Ross (CIR) assume specific functional forms for the drift and volatility of the short rate. These models are popular because they yield analytical, closed-form solutions for zero-coupon bonds.
Yield curve calibration:
- Vasicek Model: Mean-reverting drift with constant volatility, but allows negative interest rates.
- CIR Model: Mean-reverting drift with volatility proportional to $\sqrt{r}$, preventing negative rates.
- Hull-White / Ho-Lee: Incorporating time-dependent parameters to perfectly fit the current market yield curve.
9. Credit risk and default modeling require structural or intensity-based approaches.
The structural models are appealing because they are clearly closer to reality. The downside is that these models are usually more complicated to solve, with parameters that are difficult to measure.
Structural models of default. Structural models, pioneered by Merton, view a firm's equity as a call option on its total assets, with the strike price equal to the face value of its debt. Default occurs if the firm's asset value falls below its liabilities at maturity.
Reduced-form (intensity) models. Reduced-form models treat default as an exogenous, unpredictable event governed by a Poisson jump process. The hazard rate (or default intensity) can be modeled as a constant, time-dependent, or stochastic variable.
Credit risk applications:
- Credit Default Swaps (CDS): Financial contracts that transfer credit risk from the protection buyer to the seller.
- Collateralized Debt Obligations (CDOs): Structured products that pool debt and slice them into tranches with different seniorities.
- Copula Functions: Mathematical tools used to model the joint default correlation of multiple entities in a basket.
10. Portfolio management and Value at Risk quantify risk but fail during extreme market crashes.
Value at Risk is an estimate, with a given degree of confidence, of how much one can lose from one's portfolio over a given time horizon.
Modern Portfolio Theory. Markowitz's portfolio theory shows how investors can construct optimal portfolios that maximize expected return for a given level of risk (variance). This framework relies heavily on the assumption of Normally distributed asset returns.
Value at Risk (VaR). VaR is the industry standard for measuring market risk, representing the maximum expected loss over a given horizon at a specific confidence level. However, VaR assumes normal market conditions and fails to capture the extreme tail risk of market crashes.
CrashMetrics and extreme risk:
- The Failure of Normal Distributions: Real asset returns have much fatter tails than the Normal distribution predicts.
- Correlation Breakdown: During a market crash, historical correlations break down as all assets tend to fall together.
- CrashMetrics: A non-probabilistic framework that models the worst-case portfolio loss under a sudden, unhedgeable market crash.
