Need help understanding Diversification

Discussion in 'Risk Management' started by imgunn, Jul 8, 2008.

  1. imgunn


    As i understand diversification:

    1. the risk of investing in 2 or more (uncorrelated) commodities is less than the risk of investing in 1 commodity.


    0.05% risk on commodity 1
    0.05% risk on commodity 2

    ------- Less risky than --------

    0.1% risk on commodity 1

    If i understand that correctly, my question is this..

    If i maintain the same level of risk (per commodity) when adding more commodities does this increase risk and if so how ?


    0.1% risk commodity 1
    0.1% risk commodity 2

    -------- Versus --------

    0.1% risk commodity 1

    I can see how the total risk is increased... ie 0.2% at risk instead of 0.1%... BUT does that equate to increased drawdown or would the added commodities offset that ?

  2. Murray Ruggiero

    Murray Ruggiero ET Sponsor

    Simple, the theory is if you pick markets which are not highly correlated for example , gold stocks and banking stocks when one is losing you money the other is making money. Yes, they have dependency based on the market as a whole , which is why this is not perfect but it does reduce risk.
  3. Well I think you'll want to become familiar with a few concepts. You'll want to know not just how diversification effects how risky your portfolio is, but what is the best proportion of the different securities to take the least risk for a given return.

    If you assume that changes in prices are normally distributed (they aren't really but it's a useful assumption),

    - prices for a security vary and there is a standard deviation that describes how much they vary
    - prices for two securities can be correlated, this is measured by a number between -1 and 1 that describes how changes in price in one of the securities relate to changes in the other's price. 1 means they move together perfectly, -1 means that they move inversely (one up, the other down) perfectly, 0 means they aren't related
    - prices for two securities have Covariance, this equals: (Correlation of two securities) * (Standard Deviation of one security) * (Standard Deviation of the other security)

    Given these building blocks, what is the standard deviation of returns for your portfolio?

    To get the SD of the portfolio, you multiply the weight of each security (fraction of the total portfolio) by the covariance of that security with all of the other securities and then add it all up.

    If you have an idea for what the return on a position will be, you can take those into account and run a Mean-Variance Optimization on your portfolio. Here's a document that explains how to do that:

    You can do all of this in excel using some historical data and stdev() and correl() functions.