VIX 指数构造详细证明过程

VIX 指数构造详细证明过程

VIX–CBOE发行的波动率指数,又称为恐慌指数,VIX指数高的时候,说明投资者预期后市市场波动比较大,因而可以作为投资者情绪的一种预期,一般来说,如果市场是有效的话,VIX指数和标的(SPX100)之间具有杠杆效应,也就是说,标的资产的价格下降,VIX飙升。VIX指数最新的构造方法借鉴于【1】构造variance swap的方法,又称静态复制法,这篇文章的主旨是记录shecan读文章时的推导思路,所以不过多介绍背景了,直接切入正题。

VIX指数构造证明过程主要有以下的三个部分:
σ 2 = 2 T ∫ 0 T dS S − 2 T l n ( S T S 0 ) (1*) \sigma^{2} = \frac{2}{T}\int_{0}^{T}\frac{\text{dS}}{S} - \frac{2}{T}ln(\frac{S_{T}}{S_{0}}) \tag{1*} σ2=T20TSdST2ln(S0ST)(1*)

E ( σ 2 ) = 2 T ln ⁡ ( F S ∗ ) − 2 T [ F S ∗ − 1 ] + 2 T [ ∫ S ∗ ∞ 1 K 2 C T ( K ) d K + ∫ 0 S ∗ 1 K 2 P T ( K ) dK (2*) E\left( \sigma^2 \right) = \frac{2}{T}\ln\left( \frac{F}{S^{*}} \right) - \frac{2}{T}\left\lbrack \frac{F}{S^{*}} - 1 \right\rbrack \\+ \frac{2}{T}\lbrack\int_{S_{*}}^{\infty}{\frac{1}{K^{2}}C_T(K)dK +}\int_{0}^{S_{*}}{\frac{1}{K^{2}}P_T(K)\text{dK}}\tag{2*} E(σ2)=T2ln(SF)T2[SF1]+T2[SK21CT(K)dK+0SK21PT(K)dK(2*)

E ( σ 2 ) = 2 T ∑ i = 1 n Δ K i K i e rT Q ( K i ) − 1 T [ F K 0 − 1 ] 2 (3*) E(\sigma^{2}) = \frac{2}{T}\sum_{i = 1}^{n}{\frac{\mathrm{\Delta}K_{i}}{K_{i}}e^{\text{rT}}Q\left( K_{i} \right)} - \frac{1}{T}{\lbrack\frac{F}{K_{0}} - 1\rbrack}^{2}\tag{3*} E(σ2)=T2i=1nKiΔKierTQ(Ki)T1[K0F1]2(3*)

(1)根据股票过程和对数股票过程推导出波动率过程*:

假设股票过程是一个几何布朗运动,则根据伊藤引理,可以推导出对数股票过程满足的随机过程,
{ dS S = μ d t + σ d z        ( 1 ) d l n S = ( μ − 1 2 σ 2 ) d t + σ d z   ( 2 ) \left\{ \begin{matrix} \frac{\text{dS}}{S} = \mu dt + \sigma dz\ \ \ \ \ \ (1) \\ dlnS = \left( \mu - \frac{1}{2}\sigma^{2} \right)dt + \sigma dz\ (2) \\ \end{matrix} \right. {SdS=μdt+σdz      (1)dlnS=(μ21σ2)dt+σdz (2)
(1)-(2)得
1 2 σ 2 T = ∫ dS S − d l n S (3) \frac{1}{2}\sigma^{2}T =\int\frac{\text{dS}}{S} - dlnS\tag{3} 21σ2T=SdSdlnS(3)
对(3)两边求积分并化简即可得(1*)。

(2*)根据(1)构造静态复制表达式*:

首先,对(1*)两边求期望得,
E ( σ 2 ) = 2 T ( E ( ∫ 0 T dS S ) − E ( l n ( S T S 0 ) ) ) (4) E\left( \sigma^{2} \right) = \frac{2}{T}\left(E\left( \int_{0}^{T}\frac{\text{dS}}{S} \right) - E(ln(\frac{S_{T}}{S_{0}}))\right)\tag{4} E(σ2)=T2(E(0TSdS)E(ln(S0ST)))(4)
在风险中性条件下, E ( ∫ 0 T dS S ) = l n ( F S 0 ) E\left( \int_{0}^{T}\frac{\text{dS}}{S} \right) = ln(\frac{F}{S_0}) E(0TSdS)=ln(S0F), 剩下求 E ( l n ( S T S 0 ) ) E(ln(\frac{S_{T}}{S_{0}})) E(ln(S0ST)):

其中其中 S ∗ S_{*} S是静态复制时看涨期权和看跌期权的边界值,可以对 ln ⁡ ( S T S 0 ) \ln\left( \frac{S_{T}}{S_{0}} \right) ln(S0ST)做如下的分解
ln ⁡ ( S T S 0 ) = ln ⁡ ( S T S ∗ ) + ln ⁡ ( S ∗ S 0 ) (5) \ln\left( \frac{S_{T}}{S_{0}} \right) = \ln\left( \frac{S_{T}}{S_{*}} \right) + \ln\left( \frac{S_{*}}{S_{0}} \right)\tag{5} ln(S0ST)=ln(SST)+ln(S0S)(5)

最后需要证明的即是
ln ⁡ ( S T S ∗ ) = S T − S ∗ S ∗ − ∫ S ∗ ∞ 1 K 2 ( S T − K ) + d K − ∫ 0 S ∗ 1 K 2 ( K − S T ) + dK (6) \ln\left( \frac{S_{T}}{S_{*}} \right)= \frac{S_{T} - S_{*}}{S_{*}} - \int_{S_{*}}^{\infty}{\frac{1}{K^{2}}{(S_{T} - K)}^{+}dK -}\int_{0}^{S_{*}}{\frac{1}{K^{2}}\left( K - S_{T} \right)^{+}\text{dK}}\tag{6} ln(SST)=SSTSSK21(STK)+dK0SK21(KST)+dK(6)
下面给出两个证明

证明1 证明等式是否恒等

∫ 0 S ∗ 1 K 2 ( K − S T ) + dK = ∫ 0 S T 1 K 2 ( K − S T ) + dK + ∫ S T S ∗ 1 K 2 ( K − S T ) + dK = { 0                  S ∗ < K   ∫ S T S ∗ 1 K 2 ( K − S T ) + dK    S ∗ > K   \int_{0}^{S_{*}}{\frac{1}{K^{2}}\left( K - S_{T} \right)^{+}\text{dK}} =\int_{0}^{S_{T}}{\frac{1}{K^{2}}\left( K - S_{T} \right)^{+}\text{dK}}+\int_{S_{T}}^{S_{*}}{\frac{1}{K^{2}}\left( K - S_{T} \right)^{+}\text{dK}}\\= \left\{ \begin{matrix} 0\ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ S_{*} < K\ \\ \int_{S_{T}}^{S_{*}}{\frac{1}{K^{2}}\left( K - S_{T} \right)^{+}\text{dK}}\text{\ \ }S_{*} > K\ \\ \end{matrix} \right. 0SK21(KST)+dK=0STK21(KST)+dK+STSK21(KST)+dK={0                S<K STSK21(KST)+dK  S>K 

同理,有下面的结论

∫ S ∗ ∞ 1 K 2 ( S T − K ) + d K = { ∫ S ∗ S T 1 K 2 ( S T − K ) + dK    S ∗ < K   0                  S ∗ > K     \int_{S_{*}}^{\infty}{\frac{1}{K^{2}}{(S_{T} - K)}^{+}dK =}\left\{ \begin{matrix} \int_{S_{*}}^{S_{T}}{\frac{1}{K^{2}}{(S_{T} - K)}^{+}\text{dK}}\text{\ \ }S_{*} < K\ \\ 0\text{\ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ }S_{*} > K\ \\ \end{matrix} \right.\ SK21(STK)+dK={SSTK21(STK)+dK  S<K 0                S>K  

假设

{ P ( K ) = ( K − S T ) +        C ( K ) = ( S T − K ) +          \left\{ \begin{matrix} P\left( K \right) = \left( K - S_{T} \right)^{+}\text{\ \ \ \ \ }\ \\ C\left( K \right) = {(S_{T} - K)}^{+}\text{\ \ \ \ \ }\ \\ \end{matrix} \right.\ {P(K)=(KST)+      C(K)=(STK)+       

则由平价公式 S T = C T ( K ) − P T ( K ) + K S_{T} = C_{T}\left( K \right) - P_{T}\left( K \right) + K ST=CT(K)PT(K)+K,则

ln ⁡ ( S T S ∗ ) = S T − S ∗ S ∗ − ∫ S ∗ ∞ 1 K 2 ( S T − K ) + d K − ∫ 0 S ∗ 1 K 2 ( K − S T ) + dK \ln\left( \frac{S_{T}}{S_{*}} \right) = \frac{S_{T} - S_{*}}{S_{*}} - \int_{S_{*}}^{\infty}{\frac{1}{K^{2}}{(S_{T} - K)}^{+}dK -}\int_{0}^{S_{*}}{\frac{1}{K^{2}}\left( K - S_{T} \right)^{+}\text{dK}} ln(SST)=SSTSSK21(STK)+dK0SK21(KST)+dK

= S T − S ∗ S ∗ − ∫ S ∗ S T 1 K 2 ( S T − K ) + d K − ∫ S T S ∗ 1 K 2 ( K − S T ) + dK \frac{S_{T} - S_{*}}{S_{*}} - \int_{S_{*}}^{S_{T}}{\frac{1}{K^{2}}{(S_{T} - K)}^{+}dK -}\int_{S_{T}}^{S_{*}}{\frac{1}{K^{2}}\left( K - S_{T} \right)^{+}\text{dK}} SSTSSSTK21(STK)+dKSTSK21(KST)+dK

= S T − S ∗ S ∗ − ∫ S T S ∗ 1 K 2 [ ( K − S T ) + − ( S T − K ) + ] d K \frac{S_{T} - S_{*}}{S_{*}} - \int_{S_{T}}^{S_{*}}{\frac{1}{K^{2}}{\lbrack\left( K - S_{T} \right)}^{+} - \left( S_{T} - K \right)^{+}\rbrack dK} SSTSSTSK21[(KST)+(STK)+]dK

= S T − S ∗ S ∗ − ∫ S T S ∗ 1 K 2 [ S T − K ] d K \frac{S_{T} - S_{*}}{S_{*}} - \int_{S_{T}}^{S_{*}}{\frac{1}{K^{2}}{\lbrack S_T-K\rbrack dK}} SSTSSTSK21[STK]dK

= S T − S ∗ S ∗ − ln ⁡ ( S T S ∗ ) − S T S ∗ + 1 \frac{S_{T} - S_{*}}{S_{*}} - \ln\left( \frac{S_{T}}{S_{*}} \right) - \frac{S_{T}}{S_{*}} + 1 SSTSln(SST)SST+1

= ln ⁡ ( S ∗ S T ) \ln\left( \frac{S_{*}}{S_{T}} \right) ln(STS)

证明2 更加一般的形式

这个形式来源于文档【2】,当时看的时候真的觉得特别牛逼,之后很多扩展都是基于这个形式扩展的,包括skew指数的构造,但是看了作者的证明(当然作者也是特别牛逼,心生敬畏,carr and madan, 上一次看见这个名字还是在学习快速傅里叶变换应用于期权定价),云里雾里,于是按着自己的理解重新写了一遍他们的证明过程,但是数学基础还是比较欠缺,希望有缘看到我这篇笔记的大牛们来帮我订正一下~

Carr和Madan认为任何连续的payoff都可以写成如下的形式,当然个人认为这个payoff还必须是光滑的,至少要保证二阶可导,这个表达式牛逼之处就在于,后面静态复制的部分包含了所有泰勒的高阶项~
f ( S T ) = f ( S ∗ ) + f ′ ( S ∗ ) ( S T − S ∗ ) + ∫ a ∞ f ′ ′ ( K ) ( S T − K ) + d K + ∫ 0 a f ′ ′ ( K ) ( K − S T ) + dK (NB) f\left( S_{T} \right) = f\left( S_* \right) + f^{'}\left( S_* \right)\left( S_{T} - S_* \right) + \int_{a}^{\infty}{f^{''}\left( K \right)\left( S_{T} - K \right)^{+}dK +}\int_{0}^{a}{f''(K)\left( K - S_{T} \right)^{+}\text{dK}}\tag{NB} f(ST)=f(S)+f(S)(STS)+af(K)(STK)+dK+0af(K)(KST)+dK(NB)
现在我们来看看(NB)表达式的证明过程

f ( F ) = ∫ 0 ∞ f ( K ) δ ( F − K ) dK f\left( F \right) = \int_{0}^{\infty}{f\left( K \right)\delta\left( F - K \right)\text{dK}} f(F)=0f(K)δ(FK)dK

= ∫ 0 K ‾ f ( K ) δ ( F − K ) dK + ∫ k ‾ ∞ f ( K ) δ ( F − K ) dK \int_{0}^{\overline{K}}{f\left( K \right)\delta\left( F - K \right)\text{dK}} + \int_{\overline{k}}^{\infty}{f\left( K \right)\delta\left( F - K \right)\text{dK}} 0Kf(K)δ(FK)dK+kf(K)δ(FK)dK

= ∫ 0 k ‾ f ( K ) dI ( F < K ) + ∫ k ‾ ∞ f ( K ) dI ( F ≥ K ) \int_{0}^{\overline{k}}{f\left( K \right)\text{dI}\left( F < K \right)} + \int_{\overline{k}}^{\infty}{f\left( K \right)\text{dI}\left( F \geq K \right)} 0kf(K)dI(F<K)+kf(K)dI(FK)

=  f ( K ) I ( F < K ) ∣ 0 k ‾ − ∫ 0 k ‾ f ’ ( K ) I ( F < K ) dK − f ( K ) I ( F ≥ K ) ∣ k ‾ ∞ + ∫ k ‾ ∞ f ′ ( K ) I ( F ≥ K ) dK \text{\ f}\left( K \right)I\left( F < K \right)|_{0}^{\overline{k}} - \int_{0}^{\overline{k}}{f’{\left( K \right)}I\left( F < K \right)\text{dK}} - f\left( K \right)I\left( F \geq K \right)|_{\overline{k}}^{\infty} + \int_{\overline{k}}^{\infty}{f'\left( K \right)I\left( F \geq K \right)\text{dK}}  f(K)I(F<K)0k0kf(K)I(F<K)dKf(K)I(FK)k+kf(K)I(FK)dK

=  f ( k ‾ ) − ∫ 0 k ‾ f ′ ( K ) d ( K − F ) + + ∫ k ‾ ∞ f ′ ( K ) d ( F − K ) + \text{\ f}\left( \overline{k} \right) - \int_{0}^{\overline{k}}{f^{'}(K)d\left( K - F \right)^{+}} + \int_{\overline{k}}^{\infty}{f'\left( K \right)d\left( F - K \right)^{+}}  f(k)0kf(K)d(KF)++kf(K)d(FK)+

=  f ( k ‾ ) − f ′ ( K ) ( K − F ) + ∣ 0 k ‾ + ∫ 0 k ‾ f ′ ′ ( k ) ( K − F ) + dK − f ′ ( K ) ( F − K ) + ∣ k ‾ ∞ + ∫ k ‾ ∞ f ′ ′ ( K ) ( F − K ) + dK \text{\ f}\left( \overline{k} \right) - f^{'}\left( K \right)\left( K - F \right)^{+}|_{0}^{\overline{k}} + \int_{0}^{\overline{k}}{f''\left( k \right)\left( K - F \right)^{+}\text{dK}} - f'{\left( K \right)\left( F - K \right)^{+}|_{\overline{k}}^{\infty}} + \int_{\overline{k}}^{\infty}{f''\left( K \right)\left( F - K \right)^{+}\text{dK}}  f(k)f(K)(KF)+0k+0kf(k)(KF)+dKf(K)(FK)+k+kf(K)(FK)+dK

=  f ( k ‾ ) + f ′ ( k ‾ ) [ ( F − k ‾ ) + − ( k ‾ − F ) + ] + ∫ 0 k ‾ f ′ ′ ( K ) ( K − F ) + d K + ∫ k ‾ ∞ f ′ ′ ( K ) ( F − K ) + dK \text{\ f}\left( \overline{k} \right) + f^{'}\left( \overline{k} \right)\left\lbrack \left( F - \overline{k} \right)^{+} - \left( \overline{k} - F \right)^{+} \right\rbrack + \int_{0}^{\overline{k}}{f''\left( K \right)\left( K - F \right)^{+}}dK + \int_{\overline{k}}^{\infty}{f''\left( K \right)\left( F - K \right)^{+}\text{dK}}  f(k)+f(k)[(Fk)+(kF)+]+0kf(K)(KF)+dK+kf(K)(FK)+dK

f ( S T ) = ln ⁡ ( S ∗ S T ) − S T S ∗ + 1 f\left( S_{T} \right)=\ln\left( \frac{S_{*}}{S_{T}} \right) - \frac{S_{T}}{S_{*}} + 1 f(ST)=ln(STS)SST+1,则   f ′ ′ ( S T ) = − 1 S T 2 \ f^{''}\left( S_{T} \right) = - \frac{1}{{S_{T}}^{2}}  f(ST)=ST21,其中 k ˉ = S ∗ \bar{k}=S_{*} kˉ=S

代入(NB)有:
ln ⁡ ( S T S ∗ ) − S T S ∗ + 1 = − ∫ S ∗ ∞ 1 K 2 ( S T − K ) + d K − ∫ 0 S ∗ 1 K 2 ( K − S T ) + dK \ln\left( \frac{S_{T}}{S_{*}} \right) - \frac{S_{T}}{S_{*}} + 1=- \int_{S_{*}}^{\infty}{\frac{1}{K^{2}}{(S_{T} - K)}^{+}dK -}\int_{0}^{S_{*}}{\frac{1}{K^{2}}\left( K - S_{T} \right)^{+}\text{dK}} ln(SST)SST+1=SK21(STK)+dK0SK21(KST)+dK
综合(4)(5)(6)(7)可得
E ( σ 2 ) = 2 T ( E ( ∫ 0 T dS S ) − E ( l n ( S T S 0 ) ) ) = 2 T ( l n ( F S ∗ ) − F − S ∗ S ∗ + ∫ S ∗ ∞ 1 K 2 C T ( K ) d K + ∫ 0 S ∗ 1 K 2 P T ( K ) dK ) E\left( \sigma^{2} \right) = \frac{2}{T}\left(E\left( \int_{0}^{T}\frac{\text{dS}}{S} \right) - E(ln(\frac{S_{T}}{S_{0}}))\right)=\\\frac{2}{T}(ln(\frac{F}{S_*})-\frac{F - S_{*}}{S_{*}} + \int_{S_{*}}^{\infty}{\frac{1}{K^{2}}{C_T(K)dK +}\int_{0}^{S_{*}}{\frac{1}{K^{2}}P_T(K)\text{dK}})} E(σ2)=T2(E(0TSdS)E(ln(S0ST)))=T2(ln(SF)SFS+SK21CT(K)dK+0SK21PT(K)dK)
(3*) 即是对(2*)的离散过程,比较简单了,哪天心情好再更新!

【1】Demeterfi, K. , Derman, E. , Kamal, M. , & Zou, J. . (1999). More than you ever wanted to know about volatility swaps. Quantitative Strategies Research Notes Goldman Sachs.

【2】P Carr, & M Stanley. (1997). Towards a theory of volatility trading.

你可能感兴趣的:(笔记,几何学,概率论,算法)