Historical VAR

How to handle the number rounding when calculating the historical VAR?

For example, given 170 monthly returns, calculate 1% monthly VAR.

can use either 1 or 2, but since 1.7 i would giess you’d round to 2 - book says either i believe

In the book, I specifically remember that it said you can

i) Take the higher one

ii) average the two numbers

so 1.7, take either 1 or the simple average of 1 and 2

I think its a blue box somewhere.

I will add it if I can find it.

In the book, I specifically remember that it said you can

i) Take the higher one

ii) average the two numbers

so 1.7, take either 1 or the simple average of 1 and 2

I think its a blue box somewhere.

edit - its Blue box 6 page 240

I guess it should be second return ( 170*1%= 1.7%) ranked from worst to best…we choose second coz we prefer to use minimum loss at a given probability…

Thanks a lot.

its blue box# 6,p no 240

Thanks, rahuls.

Just a fun question, what if all the monthly returns are positive? How to interpret the historical VAR in this case?

It doesn’t matter whether they are positive or negative, you will still have a distribution of returns where the smallest returns will be in the lower tail (positive, negative, or zero)

When historical method could be called historical simulation method?

When we use simulated data rather than actual data points that occured in the past.

FinNinja that sounds like the argument behind National Real Estate Prices up until 2007… great example of limitation of historical VAR too.

duplicated

edit: its duplicated prev post.

If all the returns were positive - we would NOT have a VAR. Since we are talking only about Losses - when we talk VAR.

L3

When we use a different portfolio than actually had in the past to calculate historical VAR, it is called historical simulation