Serving the Quantitative Finance Community

 
User avatar
HLFX
Topic Author
Posts: 0
Joined: April 2nd, 2013, 1:59 pm

Value at Risk using implied volatolity

April 3rd, 2013, 11:10 am

A little background first. I have a portfolio of spot fx, index, equity, and commodity positions and I am attempting to build a value at risk model for the entire portfolio. I have built a very simple value at risk model in the past for a portfolio of stocks and have relied on Bloomberg for everything beyond that. Now that I have spot fx, commodity, and index positions I am struggling to create the model.I am attempting to use implied volatility as the vol input because while I am putting all this time in I want to make it a unique model. The issue I am running into using implied vol is how to derive implied correlation. In the end I want to have a 1 day 99% var on all positions and the entire portfolio. I have access to a Bloomberg Terminal, and Excel...If anyone can help I would greatly appreciate it.
 
User avatar
farmer
Posts: 61
Joined: December 16th, 2002, 7:09 am

Value at Risk using implied volatolity

April 3rd, 2013, 11:29 am

I would find the maximum empirical correlation sub-interval over the past year. Over a period of a year, just by coincidence, two assets will have a period where they move in exactly the same pattern. I would find the interval where the correlation is highest relative to the chance of it happening over that period, and use that correlation as either the maximum correlation under adverse circumstances, or even half the maximum correlation under adverse circumstances.If there were a nuclear war in Iran, I bet oil would go up, stocks would go down, and the dollar would rise at the same time. You might not see a similar cluster anywhere over your calibration interval. And you might not see it during the life of your portfolio either.
 
User avatar
farmer
Posts: 61
Joined: December 16th, 2002, 7:09 am

Value at Risk using implied volatolity

April 3rd, 2013, 11:36 am

QuoteOriginally posted by: HLFXattempting to use implied volatilityThey must have oil options in both countries denominated in local currencies, right? Maybe if the oil tails are fatter in euros than in dollars, that implies the correlation direction of dollars and euros during extreme events in that commodity.Maybe then you could curve-fit a model where the tails in equities are explained partially by volatility in commodities. Say commodities have a flat distribution, and equities have a distribution that is the sum of a normal distribution plus correlation to a portfolio of flatter distributions, including your commodities. So the correlation of equities to commodities would be implied as whatever correlation is required to get the shape of the equity implied distribution, as a sum of the normal distribution implied by nearby options plus the correlation to the implied oil distribution.
Last edited by farmer on April 2nd, 2013, 10:00 pm, edited 1 time in total.
 
User avatar
tags
Posts: 3162
Joined: February 21st, 2010, 12:58 pm

Value at Risk using implied volatolity

April 5th, 2013, 8:09 am

QuoteOriginally posted by: HLFXI am attempting to use implied volatility as the vol input because while I am putting all this time in I want to make it a unique model.That is a strange proposition. Is uniqueness your top priority?
 
User avatar
HLFX
Topic Author
Posts: 0
Joined: April 2nd, 2013, 1:59 pm

Value at Risk using implied volatolity

April 5th, 2013, 10:13 am

No uniquenss is not my top priority, however, I have read multiple sources regarding implied volatility and implied correlation and have to come the conclusion that I would at least like to test the theories myself. I believe that implied volatility could be a better indicator of realized volatility(it is a forward looking measure). Still not sure how to do this at all though. I have implied Volatility measures for each instrument, but I am not sure how to derive implied correlation from those measures.
 
User avatar
willsmith
Posts: 2
Joined: January 14th, 2008, 11:59 pm

Value at Risk using implied volatolity

April 9th, 2013, 7:58 am

Just thinking qualitatively : you can get implied vol for an instrument by looking at derivatives on that single instrument. To get implied correlation, wouldn't you need to look at derivatives priced off both instruments? Without that, I can't see how vol tells you anything about correlation.