Portfolio Standard Deviation

[mahe79]

I want to be able to calculate the standard deviation (risk) of a portfolio. The custom report has a setting were Fund Manager calculates the standard deviation of the investments in a portfolio, along with a end value weighted average for the portfolio. But I disagree that this weighted average is the correct standard deviation.

A correct portfolio std dev should be based on the investment std dev but also on the investment correlation according to my textbook. Or am I wrong??

[Mark]

Hi mahe79,

You may be right, I'm not sure of an official definition of how to calculate standard deviation for a portfolio. Do you have some reference URLs you can point me to?

[mahe79]

I searched on Google for "Calculate Portfolio Variance" and found a good URL:
http://www.oliversnotes.com/pdfs/PMT_St ... xtract.pdf

[Mark]

Hi mahe79,

Thanks for the link. I read through that document, and it does describe nicely how to calculate portfolio standard deviation. We'll shoot to implement this method in our next major update.

[mahe79]

Hi Mark,

Sounds good, I appreciate this excellent forum for posting messages!

As I see it, the portfolio risk can be calulated in two ways, either the expected risk for the future or the actual risk for a time-period in the past:

1. The method previously described used the investment correlation and volatility (std dev) along with the investment end value weight to calculate the portfolio risk. This gives opportunity to calculate the expected portfolio risk at any given date, based on historical data for the investments.

2. "Portfolio/Investment Overlay Price + Dist" gives a graph that shows how the has portfolio changed during a period. By using the data from the graph it is then possible to extract the actual portfolio std dev for the same period! The advantage is that this method takes in account if the portfolio allocation has changed during the period.

I would prefer the first method if I wanted to predict the future risk of my portfolio. But the second method is better to evaluate how a portfolio actually has fluctuated for a specific time-period. I wanted to bring this up since the second method already is partly implemented in the software, and could by easily used for this application.

[Mark]

Hi mahe79,

Thanks for the idea/feedback. One feature we're considering that would help with #2 is the ability to export graph data to a CSV file. Currently, FM can't do this.

[mahe79]

Hi Mark,

I found the answer to my original question. Calculations of the sharpe ratio uses the monthly returns of a investment or portfolio. I used the monthly returns of the portfolio from the output file "sharp_port_log.txt" and could this way manually calculate the portfolio risk (as std dev).

I can now derive the portfolio risk for any date, by simply setting FM to calculate the Sharpe ratio for that date and then use the output file!

[Mark]

Hi mahe79,

Very good. We've recently been looking into this as well, for v10. We are adding a "Standard Deviation of Monthly Returns", that does what you are doing semi-manually. That will make it easier to get the portfolio risk. This will also allow you to get risk info on other objects, like asset type, investment goal, investment type, etc...

[mahe79]

Sounds great!