How to create a curve shape histogram for a historical simulation?

Steve Jobs

Active Member
Hi,

Part 1: Assuming we want to calculate the daily VaR, assuming 252 days:
1. Get the daily P and D for the last 252 days
2. Calculate the daily R by ln[(P1+D1)/P0)]
3. Sort the data according to R
4. For 95% confidence level:
-a: 05% * 252 = 12.6 , round it to 13, add 1 to be 14, we take the highest loss no 14.
-b: or instead of 5% use the 95%, 95% * 252 = 239.4 , rounding to 239, we take the no 239 which is equal to (252-239+1=14)
5. Take the R% in no 14 from the end or the no 239 from the beginning and calculate daily VaR=R% * P252.

Part 2: In case we wanted the monthly VaR:
-a: covert the daily to monthly through the formula
-b: or, instead of daily P, get the monthly P and recalculate the VaR as above.

I hope what I mentioned in Part 1 and 2 is correct. Please advise.

But what I can't visualize is how it's possible to create a curve shaped distribution (as in below link) from a series of numbers ordered from lowest to highest which looks like a straight line?! Are the daily returns allocated to standard % ranges (bins in excel) and the same 252 returns obtained for many years?
http://www.investopedia.com/articles/04/092904.asp
 
Top