I've always been confused with Delta hedging. It is well-known that for a (smooth enough) function of (S,t) we have, due to Ito's lemma, that: dC=(∂C∂t(S,t)+μS∂C∂S(S,t)+12σ2S2∂2C∂S2(S,t))dt+σS∂C∂S(S,t)dX (source here, where X is the standard Brownian motion). The author then proceeds to show that if we choose Δ=−∂C∂S(S,t), then we will have d(C+ΔS)=(∂C∂t(S,t)+μS∂C∂S(S,t)+12σ2S2∂2C∂S2(S,t)+ΔμS)dt+σS(∂C∂S+Δ)dX=(∂C∂t(S,t)+12σ2S2∂2C∂S2(S,t))dt (I've corrected the author's typo and replaced the term ΔS(∂C∂S+Δ)dX in the formula above with the correct one σS(∂C∂S+Δ)dX )
I understand the usual procedure of Δ, but I never understand how Δ can be defined as −∂C∂S(S,t). By definition a hedging strategy must be a predictable process, is there any justification that Δ given as above is predictable, not merely adapted, to the filtration generated by the stock process?
Even if we accept the definition of Δ, the author also doesn't explain how to compute d(C+ΔS). But, as far as I know, if Δ is really the symbol for −∂C∂S(S,t), then we will have d(ΔS)=Sd(Δ)+ΔdS+(dS)⋅(dΔ). And as for dΔ, I think it's just replacing C by Δ in the expression of dC, and the resulting d(ΔS) will be super-complicated with ∂3C/∂S3 present. What I obtained in the end was a horribe mess, the risky term dX wasn't eliminated and the non-random term dt doesn't match that in the BS PDE.
Wherein lies the problem?
Answer
This question has been asked many times and some clarifications appear needed.
As pointed out in an answer to this question, the portfolio Δ1tSt+Δ2tC, where Δ1t=−∂C∂S and Δ2t=1, is, generally, neither self-financing nor locally risk-free.
To derive the Black-Scholes' PDE, we seek a portfolio Δ1tSt+Δ2tC such that it is self-financing and locally risk-free. As is shown in answer to the above question, we derived the PDE ∂C∂t+rSt∂C∂S+12∂2C∂S2σ2S2t−rC=0. and the portfolio weights Δ1t=−∂C∂SBtCt−∂C∂SS,Δ2t=BtCt−∂C∂SS, where Bt=ert is the money-market account value. Note that Δ1tSt+Δ2tC=Bt.
Based on PDE (1) and the portfolio weights give by (2), it is easy to verify that Δ1tdSt+Δ2tdC=dBt. That is, the portfolio is indeed self-financing and locally risk-free.
Though messy, based on PDE (1) and the portfolio weights give by (2), it is possible to show directly that d(Δ1tSt+Δ2tC)=Δ1tdSt+Δ2tdC using Ito's lemma. Here, we assume that the respective partial derivatives, such as ∂2C∂t∂S and ∂3C∂S3, exist.
Below, we show the self-financing property using Ito's lemma, that is, SdΔ1t+CdΔ2t+d⟨Δ1t,S⟩t+d⟨Δ2t,C⟩t=0.
For weight Δ1t, ∂Δ1t∂t=−(∂2C∂t∂SBt+r∂C∂SBt)(Ct−∂C∂SS)−(∂C∂t−∂2C∂t∂SS)∂C∂SBt(Ct−∂C∂SS)2=−∂2C∂t∂SBtCt+r∂C∂SBtCt−r(∂C∂S)2BtSt−∂C∂t∂C∂SBt(Ct−∂C∂SS)2=−∂2C∂t∂SBtCt+12∂2C∂S2∂C∂Sσ2S2tBt(Ct−∂C∂SS)2(From Eqn (1))=−∂2C∂t∂SBtC2t+12∂2C∂S2∂C∂Sσ2S2tBtCt−∂2C∂t∂S∂C∂SBtCtS−12∂2C∂S2(∂C∂S)2σ2S3tBt(Ct−∂C∂SS)3,∂Δ1t∂S=−∂2C∂S2Bt(Ct−∂C∂SS)−(∂C∂S−∂2C∂S2S−∂C∂S)∂C∂SBt(Ct−∂C∂SS)2=−∂2C∂S2BtCt(Ct−∂C∂SS)2,∂2Δ1t∂S2=−(∂3C∂S3BtCt+∂2C∂S2∂C∂SBt)(Ct−∂C∂SS)2+2(Ct−∂C∂SS)(∂2C∂S2)2SBtCt(Ct−∂C∂SS)4=−∂3C∂S3BtC2t−∂3C∂S3∂C∂SBtCtS+∂2C∂S2∂C∂SBtCt−∂2C∂S2(∂C∂S)2BtS+2(∂2C∂S2)2SBtCt(Ct−∂C∂SS)3
For weight Δ2t, ∂Δ2t∂t=rBt(Ct−∂C∂SS)−(∂C∂t−∂2C∂t∂SS)Bt(Ct−∂C∂SS)2=rBtCt−r∂C∂SBtS−Bt∂C∂t+∂2C∂t∂SSBt(Ct−∂C∂SS)2=12∂2C∂S2σ2S2tBt+∂2C∂t∂SSBt(Ct−∂C∂SS)2(From Eqn (1))=12∂2C∂S2σ2S2tBtCt+∂2C∂t∂SSBtCt−12∂2C∂S2∂C∂Sσ2S3tBt−∂2C∂t∂S∂C∂SS2Bt(Ct−∂C∂SS)2,∂Δ2t∂S=−(∂C∂S−∂2C∂S2S−∂C∂S)Bt(Ct−∂C∂SS)2=∂2C∂S2BtS(Ct−∂C∂SS)2,∂2Δ2t∂S2=(∂3C∂S3BtS+∂2C∂S2Bt)(Ct−∂C∂SS)2+2(Ct−∂C∂SS)(∂2C∂S2)2S2Bt(Ct−∂C∂SS)4=∂3C∂S3BtCtS−∂3C∂S3∂C∂SBtS2+∂2C∂S2BtCt−∂2C∂S2∂C∂SBtS+2(∂2C∂S2)2S2Bt(Ct−∂C∂SS)3.
Then d⟨Δ1t,S⟩t+d⟨Δ2t,C⟩t=(−∂2C∂S2BtCtσ2S2(Ct−∂C∂SS)2+∂2C∂S2∂C∂Sσ2BtS3(Ct−∂C∂SS)2)dt=(−∂2C∂S2BtC2tσ2S2+∂2C∂S2∂C∂SBtCtσ2S3+∂2C∂S2∂C∂Sσ2BtS3Ct−∂2C∂S2(∂C∂S)2σ2BtS4(Ct−∂C∂SS)3)dt. Moreover, SdΔ1t+CdΔ2t=S(∂Δ1t∂tdt+12∂2Δ1t∂S2σ2S2dt+∂Δ1t∂SdS)+C(∂Δ2t∂tdt+12∂2Δ2t∂S2σ2S2dt+∂Δ2t∂SdS)=S(∂Δ1t∂t+12∂2Δ1t∂S2σ2S2)dt+C(∂Δ2t∂t+12∂2Δ2t∂S2σ2S2)dt. By combining all terms together, we can show that SdΔ1t+CdΔ2t+d⟨Δ1t,S⟩t+d⟨Δ2t,C⟩t=0. Then d(Δ1tSt+Δ2tCt)=Δ1tdSt+Δ2tdCt, that is, \left(\Delta_t^1, \, \Delta_t^2\right) constitutes a self-financing strategy.
No comments:
Post a Comment