You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an alternative browser.
You should upgrade or use an alternative browser.
Does anyone have VaR
- Thread starter IV_Trader
- Start date
all you're asking and many many more is provided here. i use and i know many more that use it too. it is professional tool made available to the masses.
http://www.hoadley.net/options/options.htm
hope it helps.
http://www.hoadley.net/options/options.htm
hope it helps.
Quote from asap:
all you're asking and many many more is provided here. i use and i know many more that use it too. it is professional tool made available to the masses.
http://www.hoadley.net/options/options.htm
hope it helps.
thanks , asap. I've heard that Hoadley is very good , but I am not looking for add-ins , just for a formula.
iv
you could use the variance-covariance model developed by jp morgan (see riskmetrics.com) or monte carlo simulation. i use the latter.
the formula is pretty simple. you have to use the z score for the confidence interval you want to use i.e. 95% the z score is 1.645 times the stdev, assuming normality.
the real challenge is to model the monte carlo simulation rather than calculating var. you have to decide whether you variable follows a normal, discrete or other sort of distribution and then compute multiple simulations. each of those simulations will then be fed into a final distribution in which you'll calculate the var based on the confidence interval that suits you, using nothing more than the stdev and the appropriate z score along with the E(value).
you could use the variance-covariance model developed by jp morgan (see riskmetrics.com) or monte carlo simulation. i use the latter.
the formula is pretty simple. you have to use the z score for the confidence interval you want to use i.e. 95% the z score is 1.645 times the stdev, assuming normality.
the real challenge is to model the monte carlo simulation rather than calculating var. you have to decide whether you variable follows a normal, discrete or other sort of distribution and then compute multiple simulations. each of those simulations will then be fed into a final distribution in which you'll calculate the var based on the confidence interval that suits you, using nothing more than the stdev and the appropriate z score along with the E(value).
Quote from asap:
iv
you could use the variance-covariance model developed by jp morgan (see riskmetrics.com) or monte carlo simulation. i use the latter.
the formula is pretty simple. you have to use the z score for the confidence interval you want to use i.e. 95% the z score is 1.645 times the stdev, assuming normality.
the real challenge is to model the monte carlo simulation rather than calculating var. you have to decide whether you variable follows a normal, discrete or other sort of distribution and then compute multiple simulations. each of those simulations will then be fed into a final distribution in which you'll calculate the var based on the confidence interval that suits you, using nothing more than the stdev and the appropriate z score along with the E(value).
asap , I understand ( I think). I thought that stock's correl carries a lot of weight here. Sure , an overnight move can be easily calculate by use of SD under assumption that all stocks will move in the SAME direction under "normal" market conditions. But how to incorporate dispersion weight into overall calculations ? BTW , I am looking to evaluate buy only portfolio ( long stocks only)
I obviously need more reading on this subject.
You may want to take a look at some of the information at Gummy Stuff regarding VaR:
http://www.gummy-stuff.org/VaR.htm
There are also other financial formula and spreadsheets available.
- Greg
http://www.gummy-stuff.org/VaR.htm
There are also other financial formula and spreadsheets available.
- Greg