【FinE】期权定价理论(2) Black-Scholes模型

BS模型假设

1. 股票价格遵循几何布朗运动
2. 允许卖空衍生证券
3. 无摩擦交易环境
4. 行权期无红利支付
5. 无套利机会
6. 证券交易是连续的
7. 无风险利率是常数

BS模型推导

由假设(1)可以知道，股价$S$满足GBM方程
$$dS=\mu Sdt+\sigma Sdz$$
衍生证券价格$f$依赖于标的资产价格$S$和时间$t$，由伊藤引理得到
$$df=(\frac{\partial f}{\partial S}\mu S+\frac{\partial f}{\partial t}+\frac{1}{2}\frac{{\partial }^{2}f}{\partial S^2})dt+(\frac{\partial f}{\partial S})dz$$
联立离散形式的股价和衍生品的价格
$$\{\begin{array}{rl}& \Delta S=\mu S\Delta t+\sigma S\Delta z\\ & \Delta f=(\frac{\partial f}{\partial S}\mu S+\frac{\partial f}{\partial t}+\frac{1}{2}\frac{{\partial }^{2}f}{\partial S^2})\Delta t+(\frac{\partial f}{\partial S})\Delta z\end{array}$$
通过构造组合消除方程组中的波动项，构造1单位的衍生证券空头和$\frac{\partial f}{\partial S}$单位的股票多头，令$\Pi$表示组合价值
$$\Pi =-f+\frac{\partial f}{\partial S}S$$
在$\Delta t$时间的变化为
$$\begin{array}{rl}& \Delta \Pi =-\Delta f+\frac{\partial f}{\partial S}\Delta S\\ & \Delta \Pi =(-\frac{\partial f}{\partial t}-\frac{1}{2}\frac{{\partial }^{2}f}{\partial S^2}{\sigma }^2{S}^2)\Delta t\end{array}$$
根据价值取决于标的所承担的风险关系，$\Delta \Pi /\Pi =r\Delta t$
$$(\frac{\partial f}{\partial t}+\frac{1}{2}\frac{{\partial }^{2}f}{\partial S^2}{\sigma }^2{S}^2)\Delta t=r(f-\frac{\partial f}{\partial S})\Delta t$$
移项得到
$$\frac{\partial f}{\partial t}+\frac{1}{2}\frac{{\partial }^{2}f}{\partial S^2}{\sigma }^2{S}^2+rS\frac{\partial f}{\partial S}=rf$$

边界条件

对于欧式看涨期权，边界条件为
c ( S , t ) = max ⁡ { S − K , 0 } , t = T c(S, t)=\max\{S-K, 0\}, t=T c(S,t)=max{S−K,0},t=T
当$S(t)=0$时，期权没有价值，边界条件为
$$c(0,t)=0$$
当$S(t)\to \infty$时，$c(S,t)\to \infty$，期权的价值等于股票的价值
$$c(S,t)\sim S,S\to \infty$$

BS期权定价公式

综合BS模型和约束条件可以得到
∂ f ∂ t + r S ∂ f ∂ S + 1 2 σ 2 S 2 ∂ 2 f ∂ S 2 = r f s . t . { c ( 0 , t ) = 0 S ( t ) = 0 c ( S , t ) ≈ S , S → ∞ c ( S , T ) = max ⁡ { S − K , 0 } , t = T \frac{\partial f}{\partial t}+rS\frac{\partial f}{\partial S}+\frac{1}{2}\sigma^2S^2\frac{\partial^2f}{\partial S^2}=rf\\ s.t.\begin{cases} c(0, t)=0\\ S(t)=0\\ c(S, t)\approx S, S\to \infty \\ c(S, T)=\max\{S-K, 0\}, t=T \end{cases} ∂t∂f​+rS∂S∂f​+21​σ2S2∂S2∂2f​=rfs.t.⎩⎪⎪⎪⎨⎪⎪⎪⎧​c(0,t)=0S(t)=0c(S,t)≈S,S→∞c(S,T)=max{S−K,0},t=T​
解出欧式期权定价公式，在到期时刻为$T$，行权价格为$K$，标的资产价格$S$服从几何布朗运动的欧式看涨期权价格为
$$\begin{array}{rl}& c=SN(d_1)-K{e}^{-r(T-t)}N(d_2)\\ & d_1=\frac{\ln (S/K)+(r+{\sigma }^2/2)(T-t)}{\sigma \sqrt{T-t}}\\ & d_2=\frac{\ln (S/K)+(r-{\sigma }^2/2)(T-t)}{\sigma \sqrt{T-t}}\end{array}$$
根据平价关系可以推出欧式看跌期权的价格
$$p=K{e}^{-r(T-t)}N(-d_2)-SN(-d_1)$$

案例

$N(x)$是标准正态分布的累积分布函数，可以由以下公式近似计算
$$N(x)=\{\begin{array}{l}1-n(x)(a_{1}k+a_2{k}^2+a_3{k}^3+a_4{k}^4+a_5{k}^5),x\ge 0\\ 1-N(-x)\\ n(x)=\frac{1}{\sqrt{2\pi }}{e}^{-x^2/2}\\ k=\frac{1}{1+\gamma x}\\ \gamma =0.2316419\\ a_1=0.319381530\\ a_2=-0.356563782\\ a_3=1.781477937\\ a_4=-1.821255978\\ a_5=1.330274429\end{array}$$
例 1：计算符合标准正态分布变量$x$满足$x<1$的概率
解析：
根据近似公式计算

#include<iostream>
#include<vector>
#include<cmath>
#include<cstring>
#include<cassert>
#include<iomanip>
#include<random>
#include<map>
using namespace std;

const double pi=3.1415926;

double N(const double &x){
    if(x>6.0) return 1.0;
    if(x<-6.0) return 0.0;
    double b1=0.31938153;
    double b2=-0.356563782;
    double b3=1.781477937;
    double b4=-1.821255978;
    double b5=1.330274429;
    double p=0.2316419;
    double c=1.0/sqrt(2*pi);
    double a=fabs(x);
    double k=1.0/(1+a*p);
    double b=c*exp(-pow(x, 2)/2.0);
    double n=((((b5*k+b4)*k+b3)*k+b2)*k+b1)*k;
    n=1.0-b*n;
    if(x<0.0) n=1.0-n;
    return n;
}

int main(){
    cout<<"N(1)="<<fixed<<setprecision(4)<<N(1)<<endl;
}

python可使用scipy.stats.norm中的库函数计算

from scipy.stats import norm
print(norm.cdf(1)) # 计算累积概率密度
print(norm.ppf(0.8413)) # 计算指定cdf的分位数

例 2：假设标的资产价格为$50$，行权价格为$50$，无风险年利率是$10\%$，年波动率是$30\%$，权利期为6个月，求对应欧式看涨和欧式看跌期权价格.
解析：
计算参数
$$\begin{array}{rl}& d_1=\frac{\ln (S/K)+(r+{\sigma }^2/2)(T-t)}{\sigma \sqrt{T-t}}=\frac{(0.1+0.3^2/2)\ast 0.5}{0.3\ast \sqrt{0.5}}\\ & d_2=\frac{\ln (S/K)+(r-{\sigma }^2/2)(T-t)}{\sigma \sqrt{T-t}}=\frac{(0.1-0.3^2/2)\ast 0.5}{0.3\ast \sqrt{0.5}}\\ & c=SN(d_1)-K{e}^{r(T-t)}N(d_2)\\ & p=K{e}^{-r(T-t)}N(-d_2)-SN(-d_1)\end{array}$$

#include<iostream>
#include<vector>
#include<cmath>
#include<cstring>
#include<cassert>
#include<iomanip>
#include<random>
#include<map>
using namespace std;

const double pi=3.1415926;

double N(const double &x){
    if(x>6.0) return 1.0;
    if(x<-6.0) return 0.0;
    double b1=0.31938153;
    double b2=-0.356563782;
    double b3=1.781477937;
    double b4=-1.821255978;
    double b5=1.330274429;
    double p=0.2316419;
    double c=1.0/sqrt(2*pi);
    double a=fabs(x);
    double k=1.0/(1+a*p);
    double b=c*exp(-pow(x, 2)/2.0);
    double n=((((b5*k+b4)*k+b3)*k+b2)*k+b1)*k;
    n=1.0-b*n;
    if(x<0.0) n=1.0-n;
    return n;
}

void option_price(const double &S, const double &K, const double &r, const double &sigma, const double &tm){
    double d1=(log(S/K)+(r+pow(sigma, 2)/2)*tm)/(sigma*sqrt(tm));
    double d2=d1-sigma*sqrt(tm);
    // call option price
    double c=S*N(d1)-K*exp(-r*tm)*N(d2);
    // put option price
    double p=K*exp(-r*tm)*N(-d2)-S*N(-d1);
    cout<<"call option price: "<<c<<endl;
    cout<<"put option price: "<<p<<endl;
}


int main(){
    option_price(50, 50, 0.1, 0.3, 0.5);
    return 0;
}

python计算期权价格

from scipy.stats import norm
from math import log

def option_price(S, K, r, sigma, tm):
    d1=log(S/K)+(r+sigma**2/2)*tm/(sigma*sqrt(tm))
    d2=d1-sigma*sqrt(tm)
    c=S*norm.cdf(d1)-K*exp(-r*tm)*norm.cdf(d2)
    p=K*exp(-r*tm)*norm.cdf(-d2)-S*norm.cdf(-d1)
    print('call option price is {:.4f}'.format(c))
    print('put option price is {:.4f}'.format(p))
    
option_price(50, 50, 0.1, 0.3, 0.5)

参考资料

black-scholes model solution