I find those TV shows that talk about choosing stocks for portfolios in isolation frustrating. To me, the number of stocks, sectoral balance and funds allocated to each stock/sector are so inter-related that the whole portfolio structure should be understood before a single stock is picked. Moreover, market conditions change over time, making updates to that structure necessary at regular intervals. Changes should be considered at least once a year but ‘observations’ may be made every day to ensure the portfolio constructer hasn’t missed something!

What you need

The type of portfolio one constructs is so dependent on one’s wealth, age, health and (semi)-dependants. As I’ve said before, I have decided to transition into a predominantly yield portfolio with a little bit of growth.

But a man I met recently told me he had very little (after one of those divorces) and he was about my age. He is still working but, when he stops, he would be lucky if his money would last more than two or three years. If I were in his position, I would take on much more risk than I now intend. With good judgement and only a modicum of luck I might turn that wealth into something that would be worth preserving. I can see that other people would follow different paths.

What I do

There is an infinite number of ways a portfolio can be constructed. I use my forecasts of earnings and dividends collected by Thomson-Reuters from broking analysts and turn them into total return (including dividends) forecasts for each of the 11 major sectors of the ASX 200. These forecasts are for a rolling 12-month-ahead period and I publish them on my website in my Quant Quarterly.

I also forecast the volatility of each sector’s total returns and their correlations. I forecast these volatilities three months ahead and assume that the 12-month-ahead returns’ forecasts grow at a constant rate over the period. In that way, all of my forecasts are now for the next three months.

When I ‘optimise’ my portfolio I effectively consider every possible allocation across the sectors and choose the allocation that has the best-expected risk-adjusted return. The risk-adjusted return is the difference between the portfolio expected return and the ‘risk free rate’, which I take to be the 90-day Treasury Bill rate divided by the portfolio volatility. In actual fact, I use some heavy-handed mathematics to avert the need to consider too many possible allocations.

Without any constraints or bounds on the sector weights, some sectors might attract a negative weight. In other words, sectors with negative weights should be shorted – which means selling stocks that one doesn’t have in the hope that they can be bought back at a later date for less! Shorting is dangerous because one can actually lose more than one initially owned!

My three portfolios

So as a minimum requirement, I ensure all weights are zero or positive. I construct three types of portfolio: conviction, yield and octane. In each case, optimised weights are attached to the market index weights, which represent the relative market values of the sectors on the ASX 200. The index weights for 1 February 2014 are given in column two of the table next to the sector names.

In the high conviction portfolio, I allow sectors to be quite overweight – that is those sectors can attract more than an index-weight. I ensure that the two major sectors (materials and financials) have at least moderate (under-weight) allocations. I designed the yield portfolio to have at least index-weight for the four high yield sectors (financials, property, telcos and utilities) but the other sectors can attract a zero weight. The high-octane portfolio is the big-risk takers allocation. I show the optimised weights in columns three to six of that table.

To take the first numerical row of the table – for the energy sector – the market attracts $5.90 in a $100 portfolio. The conviction portfolio’s allocation to energy comes in at overweight ($7.90); the yield and octane portfolios each get allocated $8.80. As it happens, the energy sector has great risk-return forecasts, which take the allocations up the maxima that I choose to allow.

The differences in the weights for the materials sector across portfolio styles are even greater. The yield and octane portfolios attract no allocation to materials and the conviction has an allocation at the lower constraint than I permit.

It turns out that the weights for the financials’ sector are quite close to index weights and they are well away from my imposed constraints. Allocations depend on all of the 11 returns forecasts, 11 volatility forecasts and all of the numerous correlation forecasts.

While I wouldn’t plan on interfering with the February portfolio yet, I have shown how the optimised weights change over a one-month period in columns six to eight, which correspond to the 1 March 2014 portfolio.

The simple answer is ‘not a lot’ which is a good thing. To be considering buying and selling stocks to achieve the new allocations is not desirable in terms of transactions costs, taxes, franking credits and effort! However, in due course, weights will need to change when the underlying forecasts change.

In my next column, I will start the discussion of how to populate these sectoral allocations with stocks.

Important: This content has been prepared without taking account of the objectives, financial situation or needs of any particular individual. It does not constitute formal advice. Consider the appropriateness of the information in regards to your circumstances.

Also in the Switzer Super Report:

• Penny Pryor – Art for money’s sake
• Charlie Aitken – JB Hi-Fi best discretionary retail pick
• James Dunn – The value to be found in mining services
• Staff Reporter – Buy, Sell, Hold – what the brokers say
• Tony Negline – How to fix a problem in your SMSF
• Question of the week – SMSF versus industry funds