This Week's Random Stock List and a Response to Comment

Here's this week's random stock list with closing prices as of Friday 2/1/08. More about the experiment here.

1. Dean Foods (DF) $28.42
   
2. MoneyGram International, Inc. (MGI) $6.32
   
3. Esterline Technologies Corp. (ESL) $47.22
   
4. Fuel Systems Solutions, Inc. (FSYS) $13.03
   
5. Daktronics, Inc. (DAKT) $20.78
   
6. Stantec Inc. (SXC) $33.24
   
7. Vestin Realty Mortgage II, Inc. REIT (VRTB) $9.7
   
8. CSK Auto Corp. (CAO) $8.98
   
9. Commerce Group, Inc. (CGI) $36.25
   
10. Epicept Corp. (EPCT) $1.45
    

Curiousjoe left a comment, making the point that given that there are more small caps than other stocks in the pool from which I'm randomly selecting stocks, my picks will have a small cap bias. If I modify, or do an additional experiment where picks are market cap weighed, I'll have a large cap bias.

No matter what I do, Curiousjoe suggests, since my picks are random, their performance will mirror some market index, whether small cap, large cap, etc.
This makes for an uninteresting experiment, because whatever index the picks mirror, the performance will have to do with business cycles. Let's say the picks mirror a small cap index. The random stocks will do well in periods when small caps do well, and poorly when small caps do poorly, etc.

I agree with the above, but not entirely.

First, it's perfectly fine if the random stocks mirror an index. Indeed, they should. My hypothesis, originally stated, was "Over time stocks as a group tend to go up. You pick enough stocks, and your performance shouldn't be bad." I could have been more clear. This includes stock index performance. In fact, that's what it is.

Second, the experiment isn't about evaluating a random stock picking strategy. That is, I'm not saying "picking stocks randomly will beat the market" or anything of the sort. Most likely, as implied above, results should mirror the performance of market, or some large portion of it. The experiment's purpose is to be "one measure against which gurus and newsletters will be compared." That is, if picking stocks randomly does better than a newsletter people pay for, maybe that newsletter isn't worth it.

But here there may be the objection already mentioned: it will depend on business cycles.

I have two responses to this.

One, so what? Stock gurus and newsletters should account for business cycles in their picks. They pick stocks from all market segments, so why shouldn't I? (If certain newsletters expressly limit themselves to particular market segments I'll make sure not to use that fact against them). [For further research: are paid stock recommendations small cap biased?]

Two, to the extent the first response is inadequate, I've decided to measure the performance of the picks in a few different ways. These are as follows, and I might add more along the way depending on whether I think of something useful:

1. Same dollar amount in each stock when picked. For example, $100 per stock. Taking this week's list, for instance, this would amount to 3.519 shares of Dean Foods and 15.82 shares of MoneyGram. This will most likely have a small cap bias.

2. Dollar amount by market cap. I'd have to set the parameters, but large cap stocks would presumably affect the portfolio's performance more than small caps.

3. Averaging the performance of each stock relative to its starting price. For example, suppose I'll have 3 picks (instead of the 520 per year). Let's say one went up 50%, the other down 35%, and the last finished up 3%. Using this measure, I'll say the random picks' performance was up 6%. What bias here? Does it depend on the business cycle? While most picks will probably be small caps, there will be large and mid caps.

4. Just list how many went up and how many went down. Probably small cap bias.

I'm open to suggestions.