3 SOF Strategy Disclosed

[realcoolhead]

Me two. I found myself spending less and less time on the market, yet I am happier than ever with 3SOF. 

How often do you find someone who can change your life? I bet David is THE one!

I have come to really enjoy the 'black box' investing that 3SOF allows me.  I don't worry or spend as much time over the market anymore and that allows me concentrate on other things...like work and family.

Kudos...

[jackyw]

Hi David,

I'm new to the forum, and currently learning your 3sof model. I followed the example of AMKR and when calculating the gross margin, I found my data doesn't match. Taking FY 2004 as an example, the data shown on MSN is:

http://moneycentral.msn.com/investor/invsub/results/statemnt.aspx?Symbol=Amkr&lstStatement=Income&stmtView=Ann

Total Revenue: 1,901.28   
Gross Profit: 363.27

Gross margin = (Gross Profit)/(Total Revenue) = 19.1%

And the number in the strategy document is 31.7%

Did I miss something? Also, MSN only shows the above data (detailed income statement) up to 5 years. Any public place I could get longer history data for free?

many thanks,
Jacky

[David Randolph]

Hi David,

I'm new to the forum, and currently learning your 3sof model. I followed the example of AMKR and when calculating the gross margin, I found my data doesn't match. Taking FY 2004 as an example, the data shown on MSN is:

http://moneycentral.msn.com/investor/invsub/results/statemnt.aspx?Symbol=Amkr&lstStatement=Income&stmtView=Ann

Total Revenue: 1,901.28   
Gross Profit: 363.27

Gross margin = (Gross Profit)/(Total Revenue) = 19.1%

And the number in the strategy document is 31.7%

Did I miss something? Also, MSN only shows the above data (detailed income statement) up to 5 years. Any public place I could get longer history data for free?

many thanks,
Jacky

Excellent questions Jacky, I was waiting for someone to make that gross margin question.

I have a fundamental data service from www.alphatrade.com that costs just $5 a month (the problem is it is bundled with a $40 monthly news service ), and sometimes that service and MSN Money differ in some values (it is rare, but it happens).

I went to check the values you're telling me at the unique true source of all numbers, the SEC file 10-K for 2004, and in that case, MSN Money is in fact right, Alphatrade is wrong and the calculus for the gross margin I made for AMKR are wrong. Thanks a lot for noticing, I'll have to correct that.

To get more data for free the best place is the Yahoo Finance link "SEC filings", the problem is it will give you some work to get the data looking on all forms 10-K or 10-Q.

[jackyw]

Yep, the "SEC filings" would be the last resort unless I can't find a better source.

I can't agree more that data could differ from site to site. This time I am using ALDN.
The EPS for FY07 and 08 are:

MSN:    1.00    1.12
Yahoo: 1.17   1.30    (was 1.15    1.24  last night)

That's a big difference, but I assume the data on Yahoo is more up-to-date.

P.S. I can feel the pain of research and manually input the data after following the example. Thanks a lot for the great work!   

Jacky

[ravenquork]

Hi David,
I was looking over some of the data used in your analysis, and have noticed a little quirk in how the data is used. The outcome of the analysis is often dependent on what starting year is used. Profitable, undervalued or undiscovered companies, with good growth potential usually have a catalyst for the growth, new management, new product or even a transformation in their business model. Data prior to this catalyst is sometimes of little or no relevance to the current situation.

When you get around to further defining how your four variables are determined, would you consider looking at ways to refine how we choose basic data?  Scientist often discard outliers in their data, so why shouldn't we?     

[dnickers]

Scientist often discard outliers in their data, so why shouldn't we? 

Being a scientist, I can say that outliers are NEVER thrown out...unless you know EXACTLY why they happened (in some cases you can use another test/model to determine whether it is ok to throw them out).  But, just because they don't fit the model/hypothesis doesn't mean you can discard them.  This is a big no-no in statistics.

As for situations where a company is undergoing some change/catalyst, you're right - the model won't be accurate - but that isn't what the goal of the model is.  It uses past performance to project future values assuming certain trends. 

When a catalyst is involved, past fundamental trends mean - not so much...and estimates must be made from hearsay (think MSI - and as David has pointed out, the only one in The Red).  This increases risk, and potentially reward - but I prefer the current 3SoF approach...(great work David).

.derek

[ravenquork]

Hi Derek,

Thank you for the thoughtful comments.

Being a scientist, I can say that outliers are NEVER thrown out

Consider this – we are already throwing data out, just by picking a ten year base period, instead of 8 years or 12 years or any other arbitrary period of time.

When a catalyst is involved, past fundamental trends mean - not so much...and estimates must be made from hearsay (think MSI - and as David has pointed out, the only one in The Red).  This increases risk, and potentially reward - but I prefer the current 3SoF approach. 

I would not have picked MSI as an example of the point I was trying to make. Most of the data is suspect and the "catalyst" is actually a future event. I passed on adding that to my portfolio.

...unless you know EXACTLY why they happened (in some cases you can use another test/model to determine whether it is ok to throw them out).  But, just because they don't fit the model/hypothesis doesn't mean you can discard them.  This is a big no-no in statistics.

Well stated. It mostly, in a very lucid manner, makes the point I was trying to make. What I was suggesting was that we develop a model/hypothesis that will help us choose the yearly data set that best reflects the current operating conditions of the company.  Consider also the fact that, given the subjective nature of some of the assumptions that are used, we will never know EXACTLY what the right data set is. 

As an example, let's assume that mythical company XZX was in the asphalt shingle manufacturing business for 25 years and showed fairly steady net margins of 3.4%.  In 2003 it converted ½ of its manufacturing space to manufacturing roofing panels with integrated solar cells. Currently, (2007) they are selling everything they can manufacture at a 16% net margin. They will also complete a new plant in late 07 that will double production of the solar roofing panels.

The net margin data stream might look like this:
97   3.4;  98  3.3; 99  3.6; 00  4; 01  2.9;  02 2.2;  03 4.8; 04 7; 05 8; 06 9.6; 07E 10
a. Using data from 97 thru 06 we would arrive at an average net margin 4.9
b. Using data from 03 (when the company introduced a successful new product line) thru 06  we would arrive at an average net profit margin of 7.4
If we use a simple calculation using 1/3 of business has a net margin of 3.4; 2/3 has a net margin of 16, we might predict 08E of 11.8

I am not going to calculate CAGR for the first two models, but will leave that to any one who wants to do it. However, it is easy to intuitively see that time span selected, or methodology will make a great deal of difference in determining a final analysis. Imagine the result if our arbitrary time span included all 25 years.

but I prefer the current 3SoF approach...(great work David).

I agree great work David.
Just to be clear, I am not suggesting we throw data out just because it is inconvenient. I am not suggesting any fundamental change in strategy.

What I am suggesting is something David has talked about and that is refining the model. In this case I am suggesting that we take a look at developing a model or hypothesis that lets us decide more consistently what data is most applicable to the current condition of the company we are analyzing.
It just doesn't make sense to use older data from a company that used to make horse drawn wagons, but is know making sports cars.

Some questions we might consider:
How to determine when and how big a catalyst must be for it to influence our choice of data.
How consistent should our choice of time span be across all variables?
Should we consistently use the same time span for all analysis? Should we use something like a log scale to heavily weight the most recent data?
Which variables tend to be the most affected by choice of data set? Or are the all equal?
Should we alter the xl document to show both the arbitrary ten year results and the results from a model based the most recent business conditions?
Should we set a minimum time frame to consider, regardless of when we think a catalyst occurred?  3 years; 5 years etc.?

I am looking forward to any comments you might have. After all that is what this topic/board is about.



   

[Se7en]

David, have you ever thought about buying stocks on 'non USA markets', like the Chinese or other interesting 'booming' stock markets!?
Bad idea, why/why not? 

[Se7en]

David, have you ever thought about buying stocks on 'non USA markets', like the Chinese or other interesting 'booming' stock markets!?
Bad idea, why/why not? 

David?

[David Randolph]

David, have you ever thought about buying stocks on 'non USA markets', like the Chinese or other interesting 'booming' stock markets!?
Bad idea, why/why not? 

David?

I would like to, but access to information and trading possibilities have kept me away from that. Most people invest in foreign companies using mutual funds or hedge funds. I would like to do it directly, but I don't know where to find the information (and many times it is not in English and I don't know other languages, just Portuguese, English and Spanish) and US brokers don't trade those securities directly.

Definitely we'll get there sometime, but the world doesn't seem ready for direct foreign stock investments at this point in time. Fortunately probably the best foreign companies are also traded in the US exchanges, so we can buy their ADS's.

[tokyopua]

David, have you ever thought about buying stocks on 'non USA markets', like the Chinese or other interesting 'booming' stock markets!?
Bad idea, why/why not? 

David?

I would like to, but access to information and trading possibilities have kept me away from that. Most people invest in foreign companies using mutual funds or hedge funds. I would like to do it directly, but I don't know where to find the information (and many times it is not in English and I don't know other languages, just Portuguese, English and Spanish) and US brokers don't trade those securities directly.

Definitely we'll get there sometime, but the world doesn't seem ready for direct foreign stock investments at this point in time. Fortunately probably the best foreign companies are also traded in the US exchanges, so we can buy their ADS's.

Etrade will offer direct trading on exchanges in Germany, Canada, London, Hong Kong, Japan, and a few other countries starting in a month or two.  Canadian and London stock info should be readily availiable and in English, so you might consider that as an option.  Only problem is you have to set up cash accounts in those currencies first.

[berloga]

If USD keeps on sliding against Euro, this may be of advantage to us to trade on a foreign exchange. I am not sure how actively those exchanges, e.g. in Frankfurt am Main are traded. I know Europeans are not as actively involved in the pension and other savings through the stocks exchanges as Americans, so the question is what volumes and returns are there?

But first many of us would have to convert their USD into Euro paying a conversion fee...

[tokyopua]

If USD keeps on sliding against Euro, this may be of advantage to us to trade on a foreign exchange. I am not sure how actively those exchanges, e.g. in Frankfurt am Main are traded. I know Europeans are not as actively involved in the pension and other savings through the stocks exchanges as Americans, so the question is what volumes and returns are there?

But first many of us would have to convert their USD into Euro paying a conversion fee...

I think the conversion fee will be offset by the gain in the currency itself over the life of the investments you make as long as they are somewhat long term, i.e say half a year or longer.  Anyway, when they roll it out, I plan to give it a try and will report on the exchange rate aspects.  I will compare the rates they offer to the interbank rates reported on Oanda.com.

[David Randolph]

I'm thinking of making an important strategic change which can be summarized in two points:

1) My desire is to have a long term return on my investment of something between 30 and 40%, say 35%;

2) Yet, the valuation model only considers the stock "unattractive" when the stock, given its fundamental trends, is expected to start returning less than 16%. This means I would be satisfied with returns between 16 and 30%, which I'm not.

Therefore, should I change the "optimal selling price" calculation, from 16 to 35%? This would make the Main Portfolio much more demanding in terms of its holdings. I would have to work much harder to find an attractive candidate for the Main Portfolio.

Sounds great to me, what do you think?

[David Randolph]

The problem is my method of covering stocks with a more or less in depth video analysis wouldn't work very well with such a demanding strategy. I would have to have all fundamental data of all stocks already computed and updated with all changes to immediately spot the "35% plus" inefficiencies.

Over the next few days I'm going to search for a fundamental data service and software company that could do this for me. It will probably be a sizable investment, but I think it's worth it.

I'll keep you in touch with my findings.