One of the most misunderstood issues in trading is the notion of diversification. I can distil the standard mantra of how the sell side of the industry defines diversification into – so long as your stocks have different names you are diversified. So no need to worry.
Unfortunately, this is not true. In small closed systems such as the ASX 200 or All Ords correlation between instruments is a problem. This becomes an even bigger problem when you consider the type of stocks that are offered to most investors within a portfolio.
The table below is a correlation matrix of the ASX 20 – the cells coloured yellow represent a correlation of above 60%. Whilst those coloured blue represent a correlation of below 0%.
As you can see there are some strikingly high correlations within this closed system. There are also not as many negative correlation as you would want. The central premise regarding correlation is that it reduces risk and smooths returns. In effect it enables the investor to sleep at night.
The typical approach of a financial planner/broker would be to give an investor an equally weighted portfolio of say four stocks. This portfolio would look something like the following chart.
However, there is an assumption implicit within the concept of correlation and that is that correlations are fixed and do not drift with market conditions. If correlations changed with market conditions then this would make any investment decision based upon these correlations meaningless. And it turns out that they are.
I came across this paper – Quantifying the Behaviour of Stock Correlations Under Market Stress. It is a somewhat complicated piece and gave me a headache so I will give you the Readers Digest version. This study looked at 70 years of Dow history and ask the question – do correlations float during times of market stress?
The answer is that they do. In times of trending markets (irrespective of direction) the correlation between stocks tends to become greater. This is a problem because it is at times of market stress that investors are looking to their correlation to protect them.
The other interesting thing that the researchers found was that this drift held up over various time frames so it was fractal in nature.
As you can guess I have a few issues with diversification as it is traditionally presented. Firstly, as we have seen it is founded upon an assumption that is false. Correlations are not fixed and just when you need them to work the most they let you down. Secondly, you can only have a few good ideas for trading at any given time. I want to stick with my good ideas and only my good ideas. If gold is running I want to have as much exposure to it as my risk management allows be it via direct investment, futures, cfds, options or gold stocks.
