Why I don’t use Modern Portfolio Theory
I am a fundamental investor who believes that you must have a portfolio of stocks. My goal is to have a portfolio of 30 to 40 undervalued stocks. Unfortunately, when I tried to look for guidance on how to construct and manage such a stock portfolio, all I could find was Modern Portfolio Theory (MPT).
MPT is a statistical technique that does not work for me.
Why MPT failed
First, I am a bottom-up stock picker where I select my stocks based on value investing principles. I identify my stocks one at a time. If I start from scratch to find 30 to 40 stocks that fits my value investing requirements, it will take probably a year or two.
MPT does not help in such a case. I don’t have enough stocks to begin with for the MPT analysis to be practical. MPT is not meant for stock pickers, especially those like me who would need time to identify the 30 or 40 stocks.
Furthermore, MPT is about covariances. As a retail investor, I do not have the computing power to find the covariances for 30 to 40 stocks. On top of this, we all know that covariances are not stable. So imagine having to recompute the covariances periodically to keep track of the changes.
Just think of the dilemma. I know that individually my stocks were the best among the choices available. They were selected based on detailed company analyses with good margins of safety. But when I used MPT, the stocks will not fall on the efficient curve.
As a value investor, I believed that the market will eventually re-rate my stocks. But this will cause the future covariances to change. But the MPT computation relied in historical covariances. I did not know how to modify the historical covariances to account for the coming changes in the future.
My investment approach relies on the future being different from the past. For MPT to be useful, you have to project the future covariances. Good luck in trying to do this for 30 to 40 stocks.
But more importantly, MPT is based on the view that risk is represented by volatility. I take the view that risk is more about a permanent loss of capital rather than volatility. So I have a different risk management approach.
Over the years, I had to establish my own approach to constructing and managing a stock picking portfolio that does not rely on MPT.
My approach
I took the view that a stock portfolio is about balancing between risk and returns.
To be fair, the returns for my portfolio was estimated based on the MPT approach of the weighted expected returns of the stocks in the portfolio.
The main difference comes from how I approached risk. I took the MPT idea that for low risk, you want low correlation (this is linked to covariances) among the stocks.
But rather than try to compute the correlations, I have a set of rules of thumb to ensure that I don’t have a concentrated portfolio.
To achieve this, I sliced and diced my portfolio based on a number of characteristics e.g.
- Market cap to represent size.
- Locations. I used different stock exchanges as the proxy.
- Sectors or industries.
- Investment strategies e.g. growth stocks, cigar-butts, turnarounds.
For each characteristics, I group the stocks. The goal was that there should not be a high concentration within each characteristics. My rule of thumb was a 30 % limit for each group.
I also have a rule of thumb so that a single stock is capped at the 30% limit.
Then periodically I checked the diversity profile to ensure that they still meet my rules of thumb. Charts 1 and 2 illustrate this for the size and sector characteristics.
Has it worked? I would say that the results validate it.
If you want to know more, go to https://www.i4value.asia/2022/01/how-to-construct-winning-stock.html#more
