Boost your returns part 4 – the top 20
So far in our “Boost your Returns” series we have:
1) Introduced the idea of using a pattern recognition technology called a Support Vector Machine (SVM) to help identify listed companies with the potential to perform well/poorly (see part I here);
2) Tested the approach on historical data and found some encouraging results (see part II here); and
3) Listed the bottom 20 investment candidates from the ASX200 Index, as chosen by the machine (see part III here).
In this instalment we will move on to perhaps the most interesting part – identifying the top 20 investment candidates selected by the SVM from today’s ASX200.
Once again it’s important to note that we expect the machine to get many of its selections wrong (as would any human analyst), and the best we can expect is that it will get slightly more right than wrong. Accordingly, companies in the top 20 list should be viewed merely as candidates for further study, rather than anything more concrete.
With that behind us, the top 20 fall into line as follows:
Table 1 – Possible Outperformers
Code |
Name |
Q&P |
Sector |
CSR |
CSR LTD |
B2 |
Materials |
DOW |
DOWNER EDI LTD |
B3 |
Industrials |
TPM |
TPG TELECOM LTD |
A1 |
Telecommunication Services |
HGG |
HENDERSON GROUP PLC-CDI |
n/a |
Financials |
BGA |
BEGA CHEESE LTD |
B2 |
Consumer Staples |
MFG |
MAGELLAN FINANCIAL GROUP LTD |
A1 |
Financials |
REA |
REA GROUP LTD |
A1 |
Consumer Discretionary |
GEM |
G8 EDUCATION LTD |
A3 |
Consumer Discretionary |
LLC |
LEND LEASE GROUP |
C3 |
Financials |
CGF |
CHALLENGER LTD |
B3 |
Financials |
IAG |
INSURANCE AUSTRALIA GROUP |
A2 |
Financials |
WBC |
WESTPAC BANKING CORP |
A3 |
Financials |
DMP |
DOMINO’S PIZZA ENTERPRISES |
A2 |
Consumer Discretionary |
AAD |
ARDENT LEISURE GROUP |
B4 |
Consumer Discretionary |
RHC |
RAMSAY HEALTH CARE LTD |
B2 |
Health Care |
SGH |
SLATER & GORDON LTD |
A3 |
Consumer Discretionary |
ANZ |
AUST AND NZ BANKING GROUP |
A3 |
Financials |
MQG |
MACQUARIE GROUP LTD |
A4 |
Financials |
PPT |
PERPETUAL LTD |
B2 |
Financials |
FLT |
FLIGHT CENTRE TRAVEL GROUP |
A1 |
Consumer Discretionary |
There are some interesting names in there, some of them unsurprising, some of them perhaps a little unexpected.
Pleasingly, the SVM has put forward a relatively high-quality group of companies. Of the top 20, only 3 fall outside the A1-B3 range we require for our investment process, and these 3 are well above the bottom end of the range. Recall that when we looked at the bottom 20 chosen by the SVM we found most of the companies were outside the A1-B3 range, and many of these were at the dubious end of the scale.
A large percentage of the companies on this list are names that are very familiar to us. Four of the twenty are companies we currently own in The Montgomery Fund and/or the Montgomery [Private] Fund, and there are a number of others we have owned recently, but sold on valuation grounds, or have recently studied and rejected because of valuation.
And it is here that an interesting point starts to emerge. While we very much like a number of the companies on this list, in many cases the valuations are difficult to get comfortable with. There is a clear gap between us and the machine, and it is a gap built on investment philosophy.
Regular readers will know that our philosophy is to buy the very highest quality companies when we see them trading at a price below our estimate of their intrinsic value. When we don’t see sufficiently attractive valuations, our inclination is to hold cash and wait for better value to emerge, and that is the position we currently find ourselves in. With the market having risen faster than earnings in recent time, we have been moving more to cash to protect investor capital.
The machine has simply been asked to find the best and worst investment opportunities, regardless of whether the market as a whole represents good, bad or indifferent value, and that is what it has done. In a somewhat expensive market we shouldn’t be surprised to find it suggesting companies that we feel are expensive, notwithstanding their high quality.
In the short term, our desire to protect investor capital by holding cash is likely to hurt us in terms of investment returns. Indeed, it already has. However, in terms of long-run risk and reward, we believe it is the right thing to do, and so we let many of these opportunities go through to the keeper for now.
Time will tell if we are being too conservative in doing this. It will be interesting to keep track of the top 20 and bottom 20 companies over the next 12 months to get a sense of how the machine portfolios have performed. Of course, as with all of these sorts of exercises, we need to be careful not to read too much from a small sample size and limited time frame.
gareth hurst
:
Hello Tim,
I’m not sure if you have stated this before, but what is the investment horizon of the SVM? You stated in an earlier post that the SVM ‘liked’ momentum which may indicate a medium term <12 month investment horizon.
I would think that the investment horizon would impact the evaluation period.
Gareth
Tim Kelley
:
Hi Gareth,
We haven’t done the work to determine the best time horizon, but given the range of factors the model uses, my sense (guess) is that 12 months should be in the ballpark.
Andrew Legget
:
Thanks for the update Tim. I have found this series to be one of my favourites as i look and think more about how one can seek a greater competitive advantage in picking stocks and how things might be done better.
There is some great quality on that list and should at least help reinforce as well as the results with part 3 that quality matters. Where as i had a lapsed 1 and current none on the underperformers list i can count of at least 6 that have been mainstays of my watchlist with another 1 on the borderline. Thats encouraging for me that perhaps there is method to my own selection madness.
Looking forward to whatever future installments you have on this series.
Tim Kelley
:
Thanks for the feedback Andrew.
John Emmi
:
Hi Tim,
Thanks for the interesting articles.
There are a number SVN software implementations available. Which one you did you find the best?
Tim Kelley
:
Hi John,
Afraid we haven’t tested different implementations so can’t offer much insight on this.
johanv123
:
Two questions:
– “Pattern recognition technology”? How is this different to technical analysis and how does that fit in with value investing?
– “The machine has simply been asked to find the best and worst investment opportunities…”. Is the Skaffold “machine” not a better tool for this?
Andrew Legget
:
I think what Tim is doing above does fit into value investing. It may not seem like value investing in its purest form but what it, at least to me, is about is the objective search for anomolies or inefficiencies in the stockmarket which can be exploited for financial gain. What could actually be more value investing.
When you are in a competitive environment (such as funds management) centered around an inefficient system (stockmarket) there is an advantage in both resources (time and money spent finding the opportunities) and proft by finding out and looking deeper into potential causes of inefficiencies.
This goes even deeper for the time pressed retail investor, knowing such factors to pay attention too and those which you can discard would save invaluable time and effort.