Top Image: Segments[clusters] Formed By 4-Engine and Above ADRs
Bottom Image:Bar Chart of Each Segments Attributes [S1,S2,S3,S4]
Click on image for full-size and details.
Data and modeling is from www.valuengine.com . Self-Organizing Map and statistics uses SOMine5, courtesy of www.viscovery.net
Can ADRs be considered as a natural sector? Sectors were created based on the premise that stocks in the same sector should have some common characteristics. Or to put it another way, if we took all the attributes of each sector [based on the variables of the ValuEngine models] and did some kind of mathematical cluster analysis, distinct clusters can be seen approximating to each sector. In this sense, ADRs can viewed as a natural cluster. A Self-Organizing Map [SOM] is ideal for illustrating this. [for introduction to SOM, see http://valuengine.blogspot.com/2007/10/self-organizing-maps-som.html ] if for example, we take a list of all 5-Engine stocks together with their 40 or so attributes use a SOM to cluster the stocks based on their similarities, we would be able to discern ADRs as a distinct cluster in the SOM-space even though we did not use ADR as an attrbute. The boundaries of such a cluster would likely be more distinct at this point of time, when the gap between economic growth within and outside the U.S. for 2008 is expected to widen. We will show such a Map in a future issue of this newsletter.
For this week, we have used the new ADR screen in ValuEngine Institutional to screen for ADRs with Engine Rating of 4 or more and put them on a SOM. Then we drew a bar chart of the Valuation %, 12-month return %, Sharpe Ratio, 5-year return %, Volatility % and Expected EPS of each segment, and looked for stocks in the segment [our analysis software calls clusters "segments"] with the highest Undervaluation [measured as -%], and the highest of the other attributes i.e. with the highest +%. An explanation of the analysis is given below each graph.
For this week, we have used the new ADR screen in ValuEngine Institutional to screen for ADRs with Engine Rating of 4 or more and put them on a SOM. Then we drew a bar chart of the Valuation %, 12-month return %, Sharpe Ratio, 5-year return %, Volatility % and Expected EPS of each segment, and looked for stocks in the segment [our analysis software calls clusters "segments"] with the highest Undervaluation [measured as -%], and the highest of the other attributes i.e. with the highest +%. An explanation of the analysis is given below each graph.
Analysis: The data formed four segments, each depicted in a different color. S1= blue, S2= red, S3=yellow, S4=green. The Ticker Symbol of each is placed according to its position on the map. The physical distance between Ticker Symbols also represents the mathematical distance in the SOM space.
From the bar chart above, S4 [Segment 4] has the best overall characteristics with best Mispricing (-%) and high Last 12-month return %, Sharpe Ratio, and Volatility<--- In this case and in this market situation, Volatility was found to be highly correlated with good performance. An important observation is that these ADRs had low data values for 5-year return % and expected EPS growth. You can see that each segment is quite different from the others. Here are the stocks in Segment 4:
PHI- Philippine Long Distance Telephone Company- PHILIPPINES
HXM- Desarrolladora Homex- MEXICO
MITSY- Mitsui & Co.- JAPAN
SHG-Shinhan Financial Group- CHINA
OGZPY- Gazprom OAO- RUSSIA
BTI-British American Tobacco- UNITED KINGDOM
CEO- CNOOC Ltd- CHINA
LFC- China Life Insurance- CHINA
