To: TraderGreg who wrote (7299 ) 5/28/1998 7:47:00 PM From: jan_mike Read Replies (1) | Respond to of 11708
Stock picking contest addendum. I did some very simple number additional crunching on the contest standings. My hypothesis is that of the famous desert survival exercise where a consensus opinion is proven (usually) to be better than that of individuals. I am thinking of starting a new penny picking thread along those lines with a few wrinkles, so I wondered what the contest showed. TG says through 20 weeks the average gain is 24.5%; 22 up; 21 down; 1 unchanged. My analysis was only on those companies that were picked by more than one person, therefore presenting a consensus of sorts. My findings include the 7 companies that were picked by more than one person. SETO was picked by 5, CSMA by 8, the other 5 were picked by two each. The simple average change for these 7 companies is + 85% (I used rounded percentages in calcs) vs 24.5% for the whole sample. This is a hell of a statistic. The simple average of the two most popular picks is +22%. I didn't attempt a weighted average because of the built in CSMA bias inherent in running the contest on thread CSMA. Of the 7 multi-picked companies 4 are up and three down which is a better win/loss percentage than the entire sample too, although the small sample diminishes the significance. Of the 23 people who voted for the 7 stocks, 11 are on the plus side and 12 on the minus side. Since 8 of the 23 people were CSMA voters and presumably thread biased, a completely random contest on thread xyz would have showed even better results. Throw out, or diminish the coco biased people altogether, and the numbers really crank. Not a CSMA slander, just a statistical fact of life. Consensus picks were in fact superior, and arguably the more pickers, the righter the consensus. Respectfully submitted, Mike