二叉树模型是简化的期权定价模型,不需要复杂的随机分析
标的为股票的次日到期的看涨期权
假设条件
利率为0
符号
V V V:期权价格(二叉树模型的目标就是求出V)
S S S:今日股价
K K K:期权行权价
p p p:股价上涨概率
S u S_u Su:上涨价格
S d S_d Sd:下跌价格
期权 P a y o f f = M a x ( S − K , 0 ) Payoff=Max(S-K, 0) Payoff=Max(S−K,0)
用股票和期权构建资产无风险组合,无论价格下跌还是上涨,组合价值不发生变化
实例
S = 100 , K = 100 , p = 0.6 , S u = 101 , S d = 99 S=100, K=100, p=0.6, S_u=101, S_d=99 S=100,K=100,p=0.6,Su=101,Sd=99
组合中有一份期权,做空 N N N份股票,则
由于利率为0,三个式子应该相等,
由 1 − 101 N = − 99 N 1 - 101N = -99N 1−101N=−99N可得N=0.5,
由 V − 100 N = − 99 N V - 100N = -99N V−100N=−99N可得V=0.5
由于使用股票对期权对冲,股价上涨概率 p p p对期权价格没有影响,但是股价的波动率对期权价值有很大的影响,也就是只关心资产价格的变化幅度,不关心变化方向
为什么(二叉树模型的)理论价格是市场价格
如果不是,会产生无风险套利机会(也是理论上的),相应的买卖行为会导致市场价格回归到理论价格
股价上涨概率 p = 0.6 p=0.6 p=0.6的影响(期权Payoff的期望值为0.6):
期权的买方:
1. 大于0.6,不会有人买入
2. 0.5和0.6之间,投机者会买入,因为有正的期望收益
3. 小于0.5,不管是投机者还是对冲者都会买入
期权的卖方:
1. 大于0.5,确定会卖出
2. 小于0.5,不会卖出
对冲系数 Δ \Delta Δ
Δ \Delta Δ表示用多少份股票和一份期权可以组成无风险资产
Δ = P a y o f f ( S u , K ) − P a y o f f ( S d , K ) S u − S d \begin{aligned} \Delta=\frac{Payoff(S_u, K) - Payoff(S_d, K)}{S_u - S_d} \end{aligned} Δ=Su−SdPayoff(Su,K)−Payoff(Sd,K)
利率
( V − Δ S ) = ( P a y o f f ( S d , K ) − Δ S d ) e − r t (V - \Delta S) = (Payoff(S_d, K) - \Delta S_d)e^{-rt} (V−ΔS)=(Payoff(Sd,K)−ΔSd)e−rt
完全市场
完全市场中期权是多余的
期权、股票、现金三者中,任意两个可以复制出第三个
真实世界的特性:
风险中性世界的特性:
风险中性概率
风险中性概率是一个假想概率,使得
S = [ p ′ S u + ( 1 − p ′ ) S d ] e − r t S = [p'S_u + (1-p')S_d]e^{-rt} S=[p′Su+(1−p′)Sd]e−rt
V = [ p ′ P a y o f f ( S u , K ) + ( 1 − p ′ ) P a y o f f ( S d , K ) ] e − r t V = [p'Payoff(S_u, K) + (1-p')Payoff(S_d, K)]e^{-rt} V=[p′Payoff(Su,K)+(1−p′)Payoff(Sd,K)]e−rt
符号
S u = u S , S d = d S , 0 < d < 1 < u S_u = uS, S_d = dS, 0 \lt d \lt 1 \lt u Su=uS,Sd=dS,0<d<1<u
u u u是上涨因子
d d d是下跌因子
p p p是上涨概率
目标是通过 μ , σ \mu, \sigma μ,σ表示 u , d , p u, d, p u,d,p
均值等式
方差等式
两个方程三个未知数,有无穷多个解,选择如下近似解:
u = 1 + σ δ t u = 1 + \sigma \sqrt{\delta t} u=1+σδt
d = 1 − σ δ t d = 1 - \sigma \sqrt{\delta t} d=1−σδt
p = 0.5 + μ δ t 2 σ \begin{aligned} p = 0.5 + \frac {\mu \sqrt{\delta t}}{2 \sigma} \end{aligned} p=0.5+2σμδt
资产组合
t t t时刻资产组合
Π = V − Δ S \begin{aligned} \Pi = V - \Delta S \end{aligned} Π=V−ΔS
t + δ t t+\delta t t+δt时刻,组合的价值为
V + − Δ u S V − − Δ d S \begin{aligned} V^+-\Delta uS \ \\ \ V^- - \Delta d S \end{aligned} V+−ΔuS V−−ΔdS
由
V + − Δ u S = V − − Δ d S \begin{aligned} V^+-\Delta uS \ = \ V^- - \Delta d S \end{aligned} V+−ΔuS = V−−ΔdS
得到
Δ = V + − V − ( u − d ) S \begin{aligned} \Delta=\frac {V^+ - V^-}{(u-d)S} \end{aligned} Δ=(u−d)SV+−V−
由
V − Δ S = 1 1 + r δ t ( V + − Δ u S ) \begin{aligned} V - \Delta S = \frac{1}{1+r\delta t} (V^+-\Delta uS)\end{aligned} V−ΔS=1+rδt1(V+−ΔuS)
得到
V = V + − V − u − d + 1 1 + r δ t u V − − d V + u − d = 1 1 + r δ t 1 u − d ( ( 1 + r δ t ) ( V + − V − ) + u V − − d V + ) = 1 1 + r δ t 1 2 σ δ t [ ( 1 + r δ t ) ( V + − V − ) + ( 1 + σ δ t ) V − − ( 1 − σ δ t ) V + ) ] = 1 1 + r δ t [ ( 0.5 + r δ t 2 σ ) V + + ( 0.5 − r δ t 2 σ ) V − ] \begin{aligned} V &= \frac {V^+ - V^-}{u-d} + \frac{1}{1+r\delta t} \frac{uV^- - dV^+}{u-d} \\ &= \frac{1}{1+r\delta t} \frac{1}{u-d} ((1+r\delta t)(V^+ - V^-) + uV^- - dV^+) \\ &= \frac{1}{1+r\delta t} \frac{1}{2 \sigma \sqrt{\delta t}} [(1+r\delta t)(V^+ - V^-) \\ & \qquad + (1 + \sigma \sqrt{\delta t})V^- - (1 - \sigma \sqrt{\delta t})V^+)] \\ &= \frac{1}{1+r\delta t} [(0.5 + \frac {r \sqrt{\delta t}}{2 \sigma})V^+ + (0.5 - \frac {r \sqrt{\delta t}}{2 \sigma})V^-] \end{aligned} V=u−dV+−V−+1+rδt1u−duV−−dV+=1+rδt1u−d1((1+rδt)(V+−V−)+uV−−dV+)=1+rδt12σδt1[(1+rδt)(V+−V−)+(1+σδt)V−−(1−σδt)V+)]=1+rδt1[(0.5+2σrδt)V++(0.5−2σrδt)V−]
令
p ′ = 0.5 + r δ t 2 σ \begin{aligned} p'=0.5 + \frac {r \sqrt{\delta t}}{2 \sigma} \end{aligned} p′=0.5+2σrδt
得到
V = 1 1 + r δ t ( p ′ V + + ( 1 − p ′ ) V − ) \begin{aligned} V = \frac{1}{1+r\delta t} (p'V^+ + (1-p')V^-) \end{aligned} V=1+rδt1(p′V++(1−p′)V−)
两个概率形式一样, p p p用实际收益率 μ \mu μ, p ′ p' p′用无风险利率 r r r
p = 0.5 + μ δ t 2 σ \begin{aligned} p = 0.5 + \frac {\mu \sqrt{\delta t}}{2 \sigma} \end{aligned} p=0.5+2σμδt
p ′ = 0.5 + r δ t 2 σ \begin{aligned} p'=0.5 + \frac {r \sqrt{\delta t}}{2 \sigma} \end{aligned} p′=0.5+2σrδt
p ′ p' p′称为风险中性概率
无风险利率 r r r的两个作用:
风险中性概率和风险中性世界可以理解为一种简化建模的方法:通过调整概率,将风险厌恶程度调整为0,将收益率调整为无风险收益率r
股价二叉树:从第一期开始,往后计算,每期使用相同的参数 u , d , p u, d, p u,d,p
期权价格二叉树:计算最后一期 P a y o f f Payoff Payoff,倒数第二期的价格使用最后一期 P a y o f f Payoff Payoff的期望
连续时间 δ t → 0 \delta t \rightarrow 0 δt→0
上涨因子 u = 1 + σ δ t u = 1 + \sigma \sqrt{\delta t} u=1+σδt
下跌因子 v = 1 − σ δ t v = 1 - \sigma \sqrt{\delta t} v=1−σδt
V + = V ( u S , t + δ t ) = V ( ( 1 + σ δ t ) S , t + δ t ) ≈ V ( S , t ) + ∂ V ∂ t δ t + ∂ V ∂ S σ δ t S + 1 2 ∂ 2 V ∂ S 2 σ 2 δ t S 2 + o ( δ t ) \begin{aligned} V^+ &= V(uS, t + \delta t) = V((1 + \sigma \sqrt{\delta t})S, t + \delta t) \\ &\approx V(S, t) + \frac{\partial V}{\partial t} \delta t + \frac{\partial V}{\partial S} \sigma \sqrt{\delta t} S + \frac{1}{2} \frac{\partial^2 V}{\partial S^2} \sigma^2 \delta t S^2 + o(\delta t) \end{aligned} V+=V(uS,t+δt)=V((1+σδt)S,t+δt)≈V(S,t)+∂t∂Vδt+∂S∂VσδtS+21∂S2∂2Vσ2δtS2+o(δt)
V − = V ( v S , t + δ t ) = V ( ( 1 − σ δ t ) S , t + δ t ) ≈ V ( S , t ) + ∂ V ∂ t δ t − ∂ V ∂ S σ δ t S + 1 2 ∂ 2 V ∂ S 2 σ 2 δ t S 2 + o ( δ t ) \begin{aligned} V^- &= V(vS, t + \delta t) = V((1 - \sigma \sqrt{\delta t})S, t + \delta t) \\ &\approx V(S, t) + \frac{\partial V}{\partial t} \delta t - \frac{\partial V}{\partial S} \sigma \sqrt{\delta t} S + \frac{1}{2} \frac{\partial^2 V}{\partial S^2} \sigma^2 \delta t S^2 + o(\delta t) \end{aligned} V−=V(vS,t+δt)=V((1−σδt)S,t+δt)≈V(S,t)+∂t∂Vδt−∂S∂VσδtS+21∂S2∂2Vσ2δtS2+o(δt)
Δ = V + − V − ( u − v ) S = V ( ( 1 + σ δ t ) S , t + δ t ) − V ( ( 1 − σ δ t ) S , t + δ t ) 2 σ δ t ≈ ∂ V ∂ S \begin{aligned} \Delta &= \frac{V^+ - V^-}{(u-v)S} \\ &= \frac{V((1 + \sigma \sqrt{\delta t})S, t + \delta t) - V((1 - \sigma \sqrt{\delta t})S, t + \delta t)}{2\sigma \sqrt{\delta t}} \\ &\approx \frac{\partial V}{\partial S} \end{aligned} Δ=(u−v)SV+−V−=2σδtV((1+σδt)S,t+δt)−V((1−σδt)S,t+δt)≈∂S∂V
期权价格
V = V + − V − ( u − d ) + 1 1 + r δ t u V − − d V + u − d = 1 1 + r δ t 1 u − d [ ( 1 + r δ t ) ( V + − V − ) + u V − − d V + ) ] = 1 1 + r δ t 1 2 σ δ t [ ( 1 + r δ t ) ( V + − V − ) + ( 1 + σ δ t ) V − − ( 1 − σ δ t ) V + ) ] = 1 1 + r δ t 1 2 σ δ t [ r δ t ( V + − V − ) + σ δ t ( V + + V − ) ] = 1 1 + r δ t 1 2 σ δ t [ r δ t × 2 ∂ V ∂ S σ δ t S + σ δ t × 2 ( V + ∂ V ∂ t δ t + 1 2 ∂ 2 V ∂ S 2 σ 2 δ t S 2 ) ] = 1 1 + r δ t [ r δ t ∂ V ∂ S S + V + ∂ V ∂ t δ t + 1 2 ∂ 2 V ∂ S 2 σ 2 δ t S 2 ] \begin{aligned} V &= \frac {V^+ - V^-}{(u-d)} + \frac{1}{1+r\delta t} \frac{uV^- - dV^+}{u-d} \\ &= \frac{1}{1+r\delta t} \frac{1}{u-d} [(1+r\delta t)(V^+ - V^-) + uV^- - dV^+)] \\ &= \frac{1}{1+r\delta t} \frac{1}{2 \sigma \sqrt{\delta t}} [(1+r\delta t)(V^+ - V^-) \\ & \qquad + (1 + \sigma \sqrt{\delta t})V^- - (1 - \sigma \sqrt{\delta t})V^+)] \\ &= \frac{1}{1+r\delta t} \frac{1}{2 \sigma \sqrt{\delta t}} [r\delta t (V^+ - V^-) + \sigma \sqrt{\delta t} (V^+ + V^-)] \\ &= \frac{1}{1+r\delta t} \frac{1}{2 \sigma \sqrt{\delta t}} [r\delta t \times 2 \frac{\partial V}{\partial S} \sigma \sqrt{\delta t} S \\ & \qquad + \sigma \sqrt{\delta t} \times 2( V + \frac{\partial V}{\partial t} \delta t + \frac{1}{2} \frac{\partial^2 V}{\partial S^2} \sigma^2 \delta t S^2)] \\ &= \frac{1}{1+r\delta t} [r\delta t \frac{\partial V}{\partial S} S + V + \frac{\partial V}{\partial t} \delta t + \frac{1}{2} \frac{\partial^2 V}{\partial S^2} \sigma^2 \delta t S^2] \\ \end{aligned} V=(u−d)V+−V−+1+rδt1u−duV−−dV+=1+rδt1u−d1[(1+rδt)(V+−V−)+uV−−dV+)]=1+rδt12σδt1[(1+rδt)(V+−V−)+(1+σδt)V−−(1−σδt)V+)]=1+rδt12σδt1[rδt(V+−V−)+σδt(V++V−)]=1+rδt12σδt1[rδt×2∂S∂VσδtS+σδt×2(V+∂t∂Vδt+21∂S2∂2Vσ2δtS2)]=1+rδt1[rδt∂S∂VS+V+∂t∂Vδt+21∂S2∂2Vσ2δtS2]
移项可得到 Black-Scholes equation
∂ V ∂ t + 1 2 σ 2 S 2 ∂ 2 V ∂ S 2 + r S ∂ V ∂ S − r V = 0 \begin{aligned} \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 \end{aligned} ∂t∂V+21σ2S2∂S2∂2V+rS∂S∂V−rV=0