Saturday, September 09, 2017

Efficiently Inefficient #8 : The promised backtesting results.

As promised, I tried accessing a Bloomberg terminal to backtest some of the ideas from TUB investing. Here are my findings :

a) Dividends per unit P/B ratio was a strong screen.

Dividends generally do well in backtests. Putting up a screen incorporating the cheapness of a stock makes it even better. There is some research done this year in the Financial Analysts Journals which back-up my findings.

In fact, using Dividends divided by P/B ratio and setting it above a threshold value to target around 20 stocks yielded a return about 13.6% with a semi-variance of 13.51%. Returns were enhanced even further when I used 5 year average free cash flow yields.

TUB credited Teh Hooi Ling for this idea. This is definitely a workable investing idea.

b) ROA per unit P/B outperformed the STI ETF but the performance was unremarkable.

I did not fully adapt TUB's idea of using ROE divided by the PB ratio because ROE can be magnified by a company's gearing  and TUB used another screen to keep gearing low.  Using this screen to capture "management skill at a reasonable price" did not meet my personal expectations. Returns were at 8.86% with a semivariance of 14.19%.

c) Equity screening does not help growth investors.

This last section is something that readers should pay close attention to.

Perhaps I am still not good at equity screening, but a combination of growth screens did not yield anything concrete or useful for me. For this exercise, I searched for stocks with a combination of  high 5-year CAGR for Revenue, Zero debt, PEG ratio below 1. I gradually relaxed each criteria until I had a decent number of stocks and it returned only 3% annually over 10 years.

Because I probably fumbled with Bloomberg due to the lack of time, we should not conclude that growth investing does not work. I can only conclude that using equity screens to buy growth stocks is not a good idea.

More likely, the growth metrics were used as some sort of scorecard to funnel stocks into a bottom-up investing process that requires a more intimate investing process and an understanding of a business narrative that John happens to be quite an ace at.

Also please don't forget that Peter Lynch was a growth investor.

As of now, TUB Investing has been kind to grant me access to their database but I have not used this privilege yet because I wanted to do my own homework before using their bespoke database materials.

My simple screens, of course, do not do full justice as to the amount of work that they have done to create their database.


  1. A) do u mean returns were enhanced using dividends yield dividends by p/fcf?

  2. Yes, returns are marked improved when dividends were replaced with avg fcf yield over past 5 years.

  3. Hi Christopher, I suppose you are not a subscriber of Bloomberg terminal which I heard can cost up to 2k a month. May I know where did you get access to the terminal? SMU? Thanks

  4. SMU is one way.

    Otherwise use the central library but the wait is damn long.