Is thinking bad?

Coming back to the topic of allowing bias into the MFI methodology. I was looking at some other blogs and came across this particular one regarding FTD.

Even though it was an MFI selection, this blogger chose to avoid FTD based on a further evaluation, some of which was qualitative. The result? Assuming he had bought FTD on the date of his post and sold it exactly one year later, he would have had a 117% gain!

I don't suggest this blogger is unintelligent. Far from it, he strikes me as very intelligent. In fact, he came up with a very interesting idea. He took the principles in another value investing methodology and cross referenced it with MFI to see if he could further refine the MFI selection process.

Since these selections were made over a year ago, I decided to check the results. The average return, adjusted for splits and dividends, was 24.9% vs. S&P 500 at 21.5%, a respectable 3.5% spread over the index. Here were the results:

(split & div adjusted)
5/30/06 5/30/07 Return

AEO, 20.54, 26.98, 31.4%
BEBE, 14.11, 17.61, 24.8%
DECK, 35.22, 86.87, 146.6%
HNR, 13.74, 9.44, -31.3%
HRB, 21.9, 23.46, 7.1%
KSWS, 26.44, 28.77, 8.8%
NUE, 50.26, 67.17, 33.6%
PTEN, 28.15, 26.4, -6.2%
USNA, 37.25, 40.76, 9.4%

Avg, 24.9%

Sorry, I don't know how to make a pretty chart yet on this thing.

So, the best I could do is compare this to the MFI "fund" on those dates which had a 31.8% return. But that's not fair because the selections above are at one point in time and the MFI Fund benefits from other "tranches". So, I next compared it to my tranche from 5/8/06, which is as close as I'm going to get to those selections. My return on that tranche was 27%. And, for context, between 5/8/06 and 5/30/06, the S&P 500 declined by 5%. Therefore, the 24.9% return above was after buying into a lower market than I did at 5/8/06. (Plus, 5/7/07 to 5/30/07, the market increased 1.4%, which further strengthens the argument, but that's nominal).

Now, as I've said many times, one example does not prove anything. But, consistently, I seem to be finding that entering bias into the system hurts the results.

I'd be curious to see if anyone else has seen a system that has been resulting in better returns. I'd like to stress, I don't know that random is better...it's still too soon to tell, but I haven't found a better methodology yet.

-A

PS. Incidentally, despite my comparison above, I do find his combination of two successful methods to be an intriguing idea. I may play around with it on a mock portfolio to see what happens. Has anyone else done this?