The Motley Fool Discussion Boards

Previous Page 
Motley Fool Global Gains / GGT: Perfect Global Portfolio 

URL:
http://boards.fool.com/hiseanfirstofallthisisnotaboutyong29294769.aspx


Subject: Re: Basket cases?  Date: 5/12/2011 10:21 AM 
Author: NoFate  Number: 138 of 151 
Hi Sean First of all  this is NOT about YONG, which is why I moved the discussion into this board. This is about the general method for constructing, or general principles behind choosing to construct, a basket of stocks in an effort to attain desired risk/return profile. What my distribution of scenarios (= combination of outcome and probability) is vs. yours, vs. Tim's vs. Mr. X.' will of course be very different. The points I am making above are fairly general: 1. If you start with selecting stocks from a Group that as a whole is much higher risk, you would be inconsistent in your probability assessment (across all scenarios) if you assume that your stock picking ability moves them to a risk profile that is much lower. The risk you should attribute to the individual stock should be higher than your average because the prior information you have indicates it is much higher. This is essentially a Bayesian argument. 2. Given that the basket was constructed with a lot of weight on these stocks, it appears to me that, unfortunately, this is not what happened  i.e. the prior information (all these stocks have very high risk) was inadequately combined with the data (=stock information, familiarity, etc.) to yield a set of outcome/probability combinations that placed too much probability on positive outcomes and too little on negative ones given the prior information. This comes from Tim's figures and is what I spent so many words on in the previous posts (talking about where his probability mass seems to sit). 3. If you construct a basket with these types of stocks it seems to me you have two choices  emphasise the 'risky' one and weight it out with less risky things OR emphasise the less risky and treat the more risky as a kind of kicker. Playing around with numbers it appears to me that either of these will work the way you want it to if, and only if, the range of outcomes of the one 'thing' (here by definition the risky stock(s)) is not too wide in the context of your assessment. (By that last bit I mean that of course any stock can go to 100% but we generally assign very little probability to that and so end up effectively combining things that we assume may vary from, say, 10% to +15%.) To your questions  around Chinese microcaps  quite frankly, I do not know and anything here would be subjective. However, consider that: (a) two years ago one may reasonably said that it's probably a higher proportion than for the same market in the US 