Wednesday, June 25, 2014

Back-testing a CNAV-like portfolio using local stocks.

As a continuation to my Value Investing Master Class given to me free of charge by BigFatPurse, I spent some time in the National Library Bloomberg Terminals yesterday to see if a similar strategy may yield superior returns when applied to the local stock-market over the past 10 years.

As I'm a beneficiary of a good deed, I have to find a roundabout way to return the favor to BigFatPurse but would still benefit my readers, so I won't replicate the exact strategy shared as it would affect their business negatively, I think readers who are impressed can invest a price equal to restaurant meal for two by signing up for their course. Instead I took the broad concept to create a filter which employs a strategy similar in spirit to BigFatPurse to see market out-performance takes place.

The back-test was done over the 10 years. ( 2004 - 2014 )
The portfolio is equal-weighted.
Rebalancing of the portfolio takes place annually to reduce the impact of brokerage fees.
All companies are domiciled in Singapore to get rid of S-Chips.

My filters are as follows :

a) P/B ratio < 1
b) P/E ratio < 15
c) FCF yield > 0%
d) Debt to Equity ratio < 100%

The resulting performance was superb. About 40-50 stocks were shortlisted under this criteria.

Mean return = 26.98%
Standard Deviation = 13.53%
Sharpe Ratio = 1.29 ( Above 1 is fairly decent )

If we assume normality of returns, having a bad year with 2 standard deviations down would still return slightly below 0% so it's a pretty defensive combination of stocks.

Relative to this performance, the performance of the STI over the past 10 years was barely 7%.

Suppose when you start instituting this Graham-like strategy and it returns an extra 3% a year, with a portfolio size of $100,000, it's an extra $3,000. The model predicts an out-performance of almost 20%, making the strategy worth $20,000 a year.

This is my objective evidence to conclude that the value of the course pays itself several times over.

Disclaimer : One observation from traders is that back-tests can be overly optimistic. The likely performance of a trading strategy in practice would be below back-tested numbers but still above average market returns. If you wish to employ my screen on Bloomberg, I advise starting with a portfolio of at least $100,000 and buying at least 25 stocks to achieve diversification.


  1. Method does work. Only the Mind doesn't.

  2. Hi, the backtest shd incl S-chips since you do not know on foresight tt S-chips are bad. Second, CAGR will be a better comparison. Third, does the data used include delisted company? Fourth, if the portfolio is rebalanced say in May and not in Jan, are the returns as good?

  3. Hi ThinkNotLeft,

    I haven't seen you on online for some time. Welcome back.

    Hi Christopher,

    I agree with ThinkNotLeft that the S-chips should have been included.

    Firstly, I was one of the fools who lost a fortune in S-chips. Sigh. But I cannot verify if I am the exceptional one.

    Secondly, many value traps in the Singapore market turned out to be S-chips. On paper, their financial ratios made them look like gems in the eyes of value investors. We only know with the benefit of hindsight that they were minefields.

    Thirdly, an investor who starts out in 2004 is likely to be attracted to some of the S-chips as there was a period when they were quite hot. Even experienced value investors lost substantial money in some of them. One can read about such stories in value investment forums such as the defunct wallstraits, afralug and today's valuebuddies.

  4. The BigFatPurse philosophy actively seeks out S-Chips and discards it from the portfolio. I need to be consistent with that.

    Similarly when I construct my own portfolio, China stocks are out so it does not make sense to allow it in my tests.

    I will need to confirm whether Bloomberg covers delisted companies but I'm pretty sure their database contains de-listed stocks to eliminate survivorship bias.( Did some Googling, apparently Bloomberg draws from CRSP which minimizes survivorship bias )

    Rebalancing was set in January. I am not too sure if the rebalancing were shifted, would the results change. I have only one hour with the terminal a day and have my own strategies to backtest.

  5. Me too. Burnt a big hole with S-chips and caused portfolio CAGR to drop so much.

    Seeks out S-Chips?

    Clever trick!

    Backward looking strategy.

  6. Hi Createwealth8888,

    Thanks a lot for making me feel much better. I am not the only fool:)

    Hi Christopher,

    Thanks for the clarification.

    May I ask whether the Bloomberg terminal allows you to save the results in some external media? Do you have to book the Bloomberg terminal in advance? I am assuming you are using the one in the National library. Thanks.

  7. Hyom Hyom,

    As an unemployed uncle, we can meet for kopi then I show you my steps on the terminal.

    I have never tried saving my work but some friends say that I can ask Bloomberg staff to email the analysis even though library users are really paying customers.

  8. Hi Christopher,

    Correction. You're financially independent, not unemployed. It is one of the fewer success stories I read about engineers who made it good. Most engineer stories on is so depressing.

    Read that you are going to pursue your interest and in four years time, become a lawyer. In four years time, I will still probably be slaving as an engineer which fortunately is what I like, assuming that I still have a job.

    As for the kopi, err... sorry. I would like to keep my online identity anonymous, if you don't mind. Thanks.

  9. No problem, I respect your anonymity.

    You can always go NLB and use the EQS function in Bloomberg to filter and backtest a portfolio.

    Quite intuitive lah.

  10. Hey friends, pls don whack Chris. He did the backtest out of goodwill. Anyway, we are not trying to prove the results are so good. Personally I do not believe a return above 20% realistic. We are no warren Buffett. Our aim is to beat STI returns and if we just half the back test returns to remove the hindsight bias, survivorship bias and bid ask spread, I think that's more realistic.