📚 Core Concepts of Stochastic Models in Finance | 金融随机模型核心概念
In financial mathematics, a stochastic model is a mathematical framework that represents the random evolution of asset prices, interest rates, or other financial quantities over time. It acknowledges that future values are uncertain and models them using probability distributions and random processes.
在金融数学中，随机模型是一种数学框架，用于描述资产价格、利率或其他金融数量随时间随机演化的规律。它承认未来价值具有不确定性，并借助概率分布和随机过程来对其建模。
1. Random Variables and Financial Events | 随机变量与金融事件
A random variable assigns a numerical value to each outcome of a financial experiment. For example, the return on a stock over one day is a random variable because its exact value cannot be predicted with certainty. In A-Level mathematics, we typically work with discrete random variables (taking countable values, such as the number of defaults in a bond portfolio) and continuous random variables (taking values in an interval, such as the daily log-return of an asset).
随机变量为金融试验的每个结果赋予一个数值。例如，一只股票一天的收益率就是一个随机变量，因为我们无法确定地预测其具体取值。在A-Level数学中，我们通常处理离散随机变量（取可数个值，如债券组合中的违约次数）和连续随机变量（在某个区间内取值，如资产的日对数收益率）。
Consider a simple two-state model for a stock price. Let the current price be S₀ = £100. Over the next period, the stock can either rise to S₁ = £110 with probability p = 0.6 or fall to S₁ = £90 with probability 1 − p = 0.4. This defines a discrete random variable representing the future price.
考虑一个简单的两状态股票价格模型。设当前价格为 S₀ = 100 英镑。在下一时期，股票要么以概率 p = 0.6 上涨至 S₁ = 110 英镑，要么以概率 1 − p = 0.4 下跌至 S₁ = 90 英镑。这定义了一个表示未来价格的离散随机变量。
2. Expectation and Financial Decision-Making | 期望值与金融决策
The expected value E(X) of a random variable X is the long-term average outcome. For a discrete random variable, it is calculated as E(X) = Σ xᵢ · P(X = xᵢ). This concept is fundamental in finance because it allows investors to compare the average payoff of different investment opportunities.
随机变量 X 的期望值 E(X) 是长期平均结果。对于离散随机变量，其计算公式为 E(X) = Σ xᵢ · P(X = xᵢ)。这一概念在金融中至关重要，因为它使投资者能够比较不同投资机会的平均回报。
Using the two-state model above, the expected future price is E(S₁) = 0.6 × 110 + 0.4 × 90 = 66 + 36 = £102. If this expectation exceeds the current price plus a risk-free return, the investment may appear attractive. However, expectation alone is insufficient; risk must also be measured.
利用上述两状态模型，期望未来价格为 E(S₁) = 0.6 × 110 + 0.4 × 90 = 66 + 36 = 102 英镑。如果该期望超过当前价格加上无风险收益，则该项投资可能看起来具有吸引力。然而，仅凭期望是不够的，还必须度量风险。
The variance Var(X) = E(X²) − [E(X)]² quantifies the dispersion of possible outcomes around the mean. In finance, variance (or its square root, the standard deviation σ) serves as a measure of volatility. A higher variance implies greater uncertainty and therefore greater risk.
方差 Var(X) = E(X²) − [E(X)]² 度量可能结果围绕均值的离散程度。在金融中，方差（或其平方根，标准差 σ）作为波动率的度量。方差越大意味着不确定性越高，因此风险也越大。
E(aX + b) = aE(X) + b, Var(aX + b) = a²Var(X)
These linearity properties allow financial analysts to scale and shift positions while maintaining full tractability of the model. For a portfolio of two independent assets X and Y, the variance of the sum is Var(X + Y) = Var(X) + Var(Y).
这些线性性质允许金融分析师对头寸进行缩放和平移，同时保持模型的完全可操作性。对于由两个独立资产 X 和 Y 组成的投资组合，其和的方差为 Var(X + Y) = Var(X) + Var(Y)。
3. The Binomial Tree Model | 二项树模型
The binomial tree is a discrete-time stochastic model widely used for pricing options. It assumes that in each time step, the underlying asset price can move up by a factor u or down by a factor d. Over n time steps, the number of upward moves follows a binomial distribution B(n, p).
二项树是一种离散时间随机模型，广泛用于期权定价。它假设在每个时间步内，标的资产价格可以按因子 u 上涨或按因子 d 下跌。在 n 个时间步中，上涨次数的分布服从二项分布 B(n, p)。
After n steps, the asset price is given by Sₙ = S₀ uᵏ dⁿ⁻ᵏ, where k is the number of upward moves. The probability of exactly k upward moves is calculated using the binomial probability formula:
经过 n 步后，资产价格由 Sₙ = S₀ uᵏ dⁿ⁻ᵏ 给出，其中 k 是上涨次数。恰好发生 k 次上涨的概率通过二项概率公式计算：
P(K = k) = ⁿCₖ pᵏ (1 − p)ⁿ⁻ᵏ
For example, consider a stock with S₀ = 100, u = 1.1, d = 0.9, and p = 0.5. For a two-step tree (n = 2), a path with one up and one down move produces S₂ = 100 × 1.1 × 0.9 = £99. The probability of this particular path is 2 × 0.5 × 0.5 = 0.5, reflecting the two possible orderings of the moves.
例如，考虑一只股票，S₀ = 100，u = 1.1，d = 0.9，p = 0.5。对于一个两步树（n = 2），一次上涨加一次下跌的路径产生 S₂ = 100 × 1.1 × 0.9 = 99 英镑。该特定路径的概率为 2 × 0.5 × 0.5 = 0.5，反映了两种可能的移动次序。
4. Risk-Neutral Probabilities | 风险中性概率
In the binomial model, the actual probability p is not directly used for pricing. Instead, we compute the risk-neutral probability q = (eʳᵗ − d) / (u − d), where r is the risk-free rate and t is the length of one time step. This adjustment ensures that the expected return of the asset equals the risk-free rate, eliminating the need to estimate investor risk preferences.
在二项模型中，实际概率 p 并不直接用于定价。相反，我们计算风险中性概率 q = (eʳᵗ − d) / (u − d)，其中 r 是无风险利率，t 是单个时间步的长度。这种调整确保资产的期望收益率等于无风险利率，从而免除了估计投资者风险偏好的需要。
Suppose r = 0.05, t = 1, u = 1.1, d = 0.9. Then eʳᵗ = e⁰·⁰⁵ ≈ 1.05127, so q = (1.05127 − 0.9) / (1.1 − 0.9) = 0.15127 / 0.2 = 0.75635. The risk-neutral probability of an upward move is approximately 0.756, which is higher than the actual probability p = 0.5 because the model corrects for the upward drift required by the risk-free rate.
设 r = 0.05，t = 1，u = 1.1，d = 0.9。则 eʳᵗ = e⁰·⁰⁵ ≈ 1.05127，因此 q = (1.05127 − 0.9) / (1.1 − 0.9) = 0.15127 / 0.2 = 0.75635。上涨的风险中性概率约为 0.756，高于实际概率 p = 0.5，因为该模型对无风险利率所要求的上行漂移进行了修正。
The no-arbitrage condition requires that d < eʳᵗ < u. If this condition fails, the model permits risk-free profits, which contradicts the fundamental principle of financial markets. This condition is always checked before applying the binomial pricing framework.
无套利条件要求 d < eʳᵗ < u。如果该条件不成立，模型将允许无风险利润的存在，这与金融市场的基本原则相矛盾。在应用二项定价框架之前，必须始终检验此条件。
5. Random Walks and Brownian Motion | 随机游走与布朗运动
A random walk is a stochastic process where each successive step is determined by a random variable. In finance, the logarithm of stock prices is often modelled as a random walk with drift: ln(Sₜ) = ln(Sₜ₋₁) + μ + σεₜ, where μ is the drift, σ is the volatility, and εₜ is a standard normal random variable.
随机游走是一种随机过程，其中每一步的走向由随机变量决定。在金融中，股票价格的对数通常被建模为带漂移的随机游走：ln(Sₜ) = ln(Sₜ₋₁) + μ + σεₜ，其中 μ 是漂移项，σ 是波动率，εₜ 是标准正态随机变量。
Brownian motion (or the Wiener process) is the continuous-time limit of a random walk. A standard Brownian motion W(t) satisfies these properties: W(0) = 0; increments W(t) − W(s) are normally distributed with mean 0 and variance t − s; and increments over disjoint time intervals are independent. These properties make Brownian motion the building block of the Black-Scholes model.
布朗运动（或维纳过程）是随机游走的连续时间极限。标准布朗运动 W(t) 满足以下性质：W(0) = 0；增量 W(t) − W(s) 服从均值为 0、方差为 t − s 的正态分布；且不相交时间区间上的增量相互独立。这些性质使布朗运动成为布莱克-斯科尔斯模型的基石。
ln(Sₜ) = ln(S₀) + (μ − σ²/2)t + σ√t · Z, Z ~ N(0, 1)
This formula expresses the log-price as a normal random variable with mean ln(S₀) + (μ − σ²/2)t and variance σ²t. It enables the calculation of probabilities for various price levels at a future time, which is essential for risk assessment.
该公式将对数价格表示为均值 ln(S₀) + (μ − σ²/2)t、方差 σ²t 的正态随机变量。它使得我们能够计算未来某一时刻各种价格水平的概率，这对于风险评估至关重要。
6. The Markov Property | 马尔可夫性质
A stochastic process is said to have the Markov property if the future evolution of the process depends only on its current state, not on the entire path of past states. In mathematical form, P(Xₜ₊₁ | X₀, X₁, ..., Xₜ) = P(Xₜ₊₁ | Xₜ). This "memoryless" property is a crucial simplification in financial modelling.
如果一个随机过程未来的演化仅依赖于其当前状态，而不依赖于过去所有状态的完整路径，则该过程被称为具有马尔可夫性质。其数学形式为 P(Xₜ₊₁ | X₀, X₁, ..., Xₜ) = P(Xₜ₊₁ | Xₜ)。这种"无记忆"性质是金融建模中的关键简化。
The efficient market hypothesis suggests that all available information is already reflected in the current price of an asset. Consequently, the future price depends only on the current price and the new information that arrives, not on historical patterns. The Markov property formalises this idea.
有效市场假说认为，所有可得信息已经反映在资产的当前价格中。因此，未来价格仅取决于当前价格和即将到达的新信息，而非历史模式。马尔可夫性质将此观念形式化。
In the binomial tree model presented earlier, the probability of an upward move at any node is constant and does not depend on how the price arrived at that node. This is a direct application of the Markov property in discrete time.
在前文给出的二项树模型中，任一节点处上涨的概率是常数，不依赖于价格如何到达该节点。这是马尔可夫性质在离散时间中的一个直接应用。
7. Transition Probabilities and Markov Chains | 转移概率与马尔可夫链
A Markov chain is a discrete-state stochastic process with the Markov property. The dynamics are described by a transition matrix P, where the entry pᵢⱼ represents the probability of moving from state i to state j in one time step. Each row of the matrix sums to 1.
马尔可夫链是具有马尔可夫性质的离散状态随机过程。其动态行为由转移矩阵 P 描述，其中元素 pᵢⱼ 表示在一个时间步内从状态 i 转移到状态 j 的概率。矩阵的每一行之和等于 1。
Consider a simplified credit rating model with two states: Solvent (S) and Default (D). The transition matrix is written as:
考虑一个简化的信用评级模型，包含两个状态：偿付（S）和违约（D）。转移矩阵表示为：
| S | D | |
| S | 0.95 | 0.05 |
| D | 0.00 | 1.00 |
If there is a 0.95 probability that a solvent company remains solvent next year, and once default occurs the company remains in default with probability 1 (an absorbing state), the transition matrix above fully describes the evolution of credit quality over time. Multiplying the transition matrix by itself gives the two-year transition probabilities: P² provides the probability of moving between states over two time periods.
如果偿付公司明年仍保持偿付的概率为 0.95，而一旦违约发生，公司以概率 1 继续处于违约状态（吸收态），则上述转移矩阵完全描述了信用质量随时间的演化。将转移矩阵自乘得到两年转移概率：P² 给出在两个时间段内状态间转移的概率。
P² = P × P, and in general Pⁿ = P × P × ... × P (n times)
8. Monte Carlo Simulation | 蒙特卡洛模拟
Monte Carlo simulation is a computational technique that uses repeated random sampling to estimate the distribution of possible outcomes. In finance, it is used to price complex derivatives and to assess portfolio risk when closed-form formulas are unavailable. The method is built upon the law of large numbers: as the number of simulated paths N increases, the average simulated payoff converges to the true expected payoff.
蒙特卡洛模拟是一种利用重复随机抽样来估计可能结果分布的计算技术。在金融中，当封闭形式的公式不可用时，它用于复杂衍生品定价和投资组合风险评估。该方法以大数定律为基础：随着模拟路径数量 N 的增加，模拟平均收益收敛于真实期望收益。
For the random walk model ln(Sₜ) = ln(S₀) + (μ − σ²/2)t + σ√t · Z, the simulation procedure is straightforward. For each simulated path, draw a random value of Z from the standard normal distribution, compute the future price Sₜ, and record the payoff of the derivative. After generating N = 10,000 paths, the average payoff is discounted at the risk-free rate to obtain the derivative price.
对于随机游走模型 ln(Sₜ) = ln(S₀) + (μ − σ²/2)t + σ√t · Z，模拟过程是直接的。对于每条模拟路径，从标准正态分布中抽取随机值 Z，计算未来价格 Sₜ，并记录衍生品的收益。在生成 N = 10,000 条路径后，以无风险利率对平均收益进行贴现，即可得到衍生品价格。
The standard error of the Monte Carlo estimate decreases as 1/√N. Therefore, quadrupling the number of simulated paths halves the estimation error. This trade-off between computational cost and accuracy is fundamental in applied quantitative finance.
蒙特卡洛估计的标准误差按 1/√N 的速度递减。因此，将模拟路径数量扩大四倍可以使估计误差减半。计算成本与精度之间的这种权衡在应用定量金融中至关重要。
9. Applications and Limitations | 应用与局限
Stochastic models are applied extensively in insurance, investment banking, and corporate finance. Insurance companies use them to model claim frequencies and severities. Investment banks use them to price options and to manage market risk. Corporate treasurers use them to evaluate the risk of interest rate movements on debt portfolios.
随机模型在保险、投资银行和公司金融中有着广泛的应用。保险公司使用它们来模拟理赔频率和损失程度。投资银行用它们来为期权定价和管理市场风险。公司财务主管用它们来评估利率变动对债务组合的风险。
The binomial tree model is intuitive and flexible, making it ideal for valuing American options where early exercise is permitted. Brownian motion-based models are compact and analytically tractable, but they assume normally distributed returns and constant volatility. In reality, financial returns exhibit fat tails (extreme events occur more frequently than the normal distribution predicts) and volatility clustering (periods of high and low volatility alternate). Model risk arises precisely because these assumptions may fail under extreme market conditions.
二项树模型直观灵活，非常适合对允许提前行使的美式期权进行估值。基于布朗运动的模型紧凑且解析可处理，但它们假设收益率服从正态分布且波动率恒定。现实中，金融收益表现出肥尾特征（极端事件比正态分布预测的更频繁发生）和波动率聚集现象（高波动期与低波动期交替出现）。模型风险恰恰源于这些假设在极端市场条件下可能失效。
Despite these limitations, stochastic models remain indispensable tools. Their value lies not in perfect prediction but in providing a structured, quantitative framework for understanding uncertainty, comparing alternatives, and communicating risk in a rigorous manner.
尽管存在这些局限性，随机模型仍然不可或缺。它们的价值不在于完美预测，而在于提供一个结构化的量化框架来理解不确定性、比较不同方案并以严谨的方式沟通风险。
10. Conclusion | 总结
Stochastic models provide the mathematical language for describing financial uncertainty. From discrete binomial trees to continuous Brownian motion, these frameworks allow us to quantify expected returns, measure risk, and price derivative securities. The Markov property and the concept of risk-neutral probabilities are pivotal ideas that link probability theory to financial economics.
随机模型提供了描述金融不确定性的数学语言。从离散的二项树到连续的布朗运动，这些框架使我们能够量化期望收益、度量风险并定价衍生证券。马尔可夫性质与风险中性概率概念是将概率论与金融经济学联系起来的关键思想。
Mastery of these core concepts—random variables, expectation, the binomial tree, and the normal distribution—equips students with the analytical tools needed to understand more advanced models and their real-world applications. As markets evolve and data becomes richer, the principles of stochastic modelling will continue to underpin modern financial practice.
掌握这些核心概念——随机变量、期望、二项树和正态分布——赋予学生理解更高级模型及其现实应用所需的分析工具。随着市场不断演进、数据日益丰富，随机建模的原理将持续支撑现代金融实践。
Published by TutorHao | Mathematics Revision Series | aleveler.com
更多咨询请联系16621398022（同微信）
或截图后用「扫一扫 → 相册」选择本图