During World War II, natives living in the islands involved in the War of the Pacific saw soldiers call upon metallic birds which flew past and dropped tin cans which contained delicious food. After the war, a cargo cult was born - natives built planes out of straw to attract these metallic birds which they hoped would drop more food on them.
Some people ask me why I am so obsessed with results rather than process. The reason is that I do not wish to fall into the same category as these cargo cultists. Process is important because it's a shortcut to consistent results, but if I can't get my results, the process has to be improved further. This is even if the process has been adhered perfectly.
The book uses an example of an author who wants to write a book. Instead of writing a book, he goes on to emulate the lifestyle of Ernest Hemingway, living a playboy lifestyle and using only Moleskines notebooks to capture his literary ideas. His book eventually flopped.
To be fair, my investment approach is not immune to the problem of cargo cultism.
When you keep using quantitative backtesting tools to create a high-return low-risk strategy, you can end up using a combination of factors to data-mine historical results. If you follow this strategy correctly, it may not translate to future earnings.
My defence against data mining is to use only factors suggested by quantitative investors who have combed US data for 50+ years. All I do is employ those same factors that work in the US to Singapore Stock Exchange for the past decade with some adaptations of my own.
Here's an example of what I am working on next for my next workshop :
If you are familiar with the characteristics of the STI components, you will be aware that returns have been about 7+% at most for the past decade. Utilising factors in investing does not significantly boost returns because these are stocks that are monitored by major investment houses so markets are fairly efficient. This is frustrating given that STI components attract cheap financing for leveraged portfolios.
The question is whether is it possible to create a different kind of STI the same way US quants call Market Leaders. Market leaders are large, successful businesses that brokers would be keen to provide cheap margin for.
So guided by the existing literature on the US markets, I developed a localised form of Market Leader stocks.
- Stocks must be found in SGX
- Filter out China domiciled counters and REITs.
- Stocks must have a market capitalisation above $200M for adequate liquidity.
- Stocks must be above the median in free cash flow.
- Stocks must be above the median in revenues.
This set of filters will generate about 40+ counters which generate 15.77% returns with a low semi-variance of 11.81% (15 year history plus annual rebalancing) making this a superior starting point for investors to begin quantitatively selecting factors to achieve superior performance.
Of course this on only a teaser, follow up considerations would be adequately considered in the second run of my workshop.