Steven Shreve Bk 2 Pg 467/468 Topic Martingale property

Dear experts,

Can anyone please help me clarify the below topic. I have searched the net but could not find any/meaningful explanation

M(t)Represents compensated Poisson property as martingale property
N(t)The Poisson process N(t) counts the number of jumps that occur at or before time t.
λtThe Poisson process N(t) has intensity λ
‘λt is the Expected value of the Poisson process in interval time [0,t].


Compensated Poisson process is defined as M(t) = N(t) – λt. Then M(t), Compensated Poisson process is a martingale.

Usually, the Poisson process has lines parallel to time (X axis) to show time until the next jump.

I understand the proof but not the above diagram.

Question1: What does straight line with downward slope mean? And what do they characterise?

Question2: Why is it important for me to understand that the compensated Poisson process is a Martingale
 

Attachments

  • Steven Shreve Bk 2 Pg 467 Topic Martingale property.docx
    31.4 KB · Views: 10
What does straight line with downward slope mean? And what do they characterise?
That is your -\lambda. It is downward sloping.

Why is it important for me to understand that the compensated Poisson process is a Martingale
The "compensation" comes up after you apply the change of measure ( i called it lambda over lambda, to be exact, lambda_tilt / lambda ). Then your discounted*S will become a martingale. This is the master equation you need to remember for change of measure under jump diffusion,

[math]dS=\mu S dt + \sigma S dW_t + S(t-)dQ_t[/math][math]Z_1=\exp {\int_0^t \theta dW_s - \frac{1}{2} \theta^2ds }[/math][math]Z_2=\exp{(\lambda - \lambda')t } \prod_s (\frac{\lambda'}{\lambda})^{N(s)}[/math][math]Z=Z_1Z_2[/math]
Then you have,
[math]dS=rSdt+\sigma S dW_t^Q+S(t-)dM_t^Q = ( r-\lambda \mu )Sdt+\sigma S dW_t^Q+S(t-)dQ_t^Q[/math]
 
Thank you PepeQuant for the prompt reply.

My Phd. sir will be coming in the evening. I will show your response to him and if need I will comeback with further doubts/clarifications.

I will be raising few more questions today. If your time permits, kindly take at them.

Thank you once again, a lot. Every small help means a lot to my anxiety to understand this subject.

My email id is kotadp-mfe@yahoo.com
 
Hello Pepe,

Thank you for your response.

My understanding is lambda is the intensity of the Poisson process. Since Poisson is a counting process, lambda has to be a positive integer.

Here how does lambda take a negative value?

Is the downward slope because of the negative value?
 
You have listed Z1, Z2 and Z formulae. These are from chapter 11, equation 11.6.31/32.33, I think, thank you.

My Tutor is unclear about the derivation of the below differential of the stock process:

dS=rSdt+σSdWtQ+S(t−)dMtQ=(rλμ)Sdt+σSdWtQ+S(t−)dQtQ

Are you able to provide me further information how the above differential was arrived using the previous information

Thank you
 
Last edited:
Hello Pepe,

Thank you for your response.

My understanding is lambda is the intensity of the Poisson process. Since Poisson is a counting process, lambda has to be a positive integer.

Here how does lambda take a negative value?

Is the downward slope because of the negative value?
I think you are misunderstanding the construction of the Poisson/compound Poisson process and should consider re-reading the first few sections of Ch. 11 since the rest of the derivations in the chapter will never “click” if you don’t understand the basics. Lambda does not need to be a positive integer.
 
Hi Qui-Gon/PepeQuant.

In all honesty, I have struggled to understand the construction of Random variable Y,y compound Poisson Districtuion.

But here is the bare truth, my PhD maths sir has managed to give me notes almost for every other section/point in detail in this chapter. My detailed notes on topics where we could, for this chapter alone, has crossed 300 pages. I am glad I could write detailed notes where I could. I wish I can write more.

We also struggled to write proof for Theorem 11.4.5 Pg 477 (Steven Shreve's Stochastic Calculus book 2).

He has worked hard and we could not manage to get any notes on these topics.

Hence we are pleading and kneeling for guidance/pointers from the experts in this forum.

I will raise another question/thread for theorem 11.4.5

Thank you all
 
Top