Monte Carlo simulation question

[Fra_maf]

Hi everyone,

I would like to ask you some clarification about the implementation of MC simulation.

Just to clarify:

Nsim = number of simulation, Nsteps = number of steps


I get that if I increase Nsim I reduce the confidence interval and therefore I obtain a "more precise result".


If I increase Nsteps, I would say that I get a better estimation of the price itself. In the sense that for example if I want to compute the price of a plain vanilla call on stock which follow the GBM dynamics, the more I increase the number of steps the more my estimated price will get closer to the B&S formula price.


Is my idea correct or am I missing something?


Thank you and sorry for the (probably) stupid question

 

[mhy]

Yes, if you increase the number of steps (i.e. decrease your estimate of 'dt') you will get closer and closer to the continuous case and will get closer to the B/S price.

 

[John TRD]

Nsim = number of simulation, Nsteps = number of steps

Thank you and sorry for the (probably) stupid question

Hi looks like i have simillar question Which monte carlo implementation do you using& can y give me link please

 

[Daniel Duffy]

"I would like to ask you some clarification about the implementation of MC simulation."

MC simulation of what?

You are describing the solution to a problem, but I think you have forgotten to tell us what the problem is.

 

[Daniel Duffy]

Hi looks like i have simillar question Which monte carlo implementation do you using& can y give me link please

You have lots of questions But what's your problem ?

 

[John TRD]

You have lots of questions But what's your problem ?

Problem it is backtester for stock quotes. (So ihave function which calculate profit on historic prices with trading rules sell if price > A buy if price < B Hold if price=C A,B,C is parameters for optimisation/ And i sampling them via monte carlo )
Monte carlo have 2 options Runs anf Test and Runs algorithm will find first local maximum and another sample will be near this range/ For example in my stoc testter fist sample A, B random vaiable s with range (1..20)
first sample found local maximum (A,B)A=4 B=6) And next sampling will be near this values A+-3 B+-3 so ranndom variable will be from range A1 (1,2,3,4,5,6,7) B1(3,4,5,6,7,8,9) IT IS RUN ALGORITHM
TEST algrithm will use all range A(1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20) and B (1-20)

And i can set manually before running optimisation how much Runs and test will be Runs 5 and Test 10 mean After each Run will be 10 TESst and Total RUN*Test=50 samples

So i should find simillar monte carlo model/Actualy i dont know how thsi can be imlementet (i should i normal distribution to fin random variable in theese ranges)And i don t know which ranges should be and which probablity/How strong next samples should be centered near local maximum

I am looking for papers,articles about this//How this Runs -Test procedure calling in statistics? Resampling?

Lets talk in another thread