Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Pastimes : Q&A on SI's New Feature: Investor Sentiment -- Ignore unavailable to you. Want to Upgrade?

To: Edscharp who wrote (46)	1/10/2003 3:35:06 AM
From: Chief AllPredictive	Respond to of 82

Ed, I think it is more than just an entertainment. But we also don't know exactly how the correlations will play out and we are open about it. Until we say otherwise, the conclusions are the responsibility of the user. (Have you read our Disclaimer?) Investing is the art of skillfully using all data available. If you use the sentiment right - confirm with fundamentals and maybe even technical analysis, then it may work for you. If you do technical analysis, you know that changing parameters for an indicator may change its buy signal to a sell signal or vice versa. The fundamental research certainly did not work for many in the near past. Accounting choices always highlight one aspect of the business and hide another one. Regardless of what the politicians say, no reform will make it go away. Investment pros I spoke with expect investor sentiment to be a contrarian indicator and are interested in the data. I understand that your position is that they are wrong because of all the reasons you posted. If they read this posting, they will feel ashamed that they wanted the data and will not call me again. Why should I defend them? At least you see that even some pros are messed up with this sentiment thing. :-) Now seriously. It seems that what you want is to have a serious discussion on statistics and sampling theory. That may be more than this thread can bear (and enjoy) but let me at least try to make a few points: 1. Obtain predictions from smart pros instead of dumb amateurs There is a whole theory about how a lot of "amateurs" can outsmart a few "pros". Not individually but in an aggregation. The point is that the amateurs know "something" about the topic and there are many of them. That said, I expect also some pros to join the service. When we track the accuracy of predictions, we do not know if you are a billion dollar money manager or an investor with IRA. But we will know how accurate your predictions are and we can use that info to make published predictions better. How does a farmer in South Dakota or a policeman in Wisconsin know what the US “national” GDP will be in the next 6 or 12 month? The fact is, they don't. They are all wrong. They experience only their local economy. But when you make an average, it comes out surprisingly right, month after month, year after year (or at least better than what pros at FED and major banks produce). (I am talking about Michigan Consumer Survey of 500 randomly selected Americans.) I think these economists are sincere, extremely smart people with all the data and analytical tools that can be. It is just an example of many "amateurs" outsmarting a few "pros". 2. Random versus self-selected samples If a researcher can select a random sample of knowledgeable people in a cost-effective and timely manner, s/he will. The fact is, most surveys in the real world are not using random samples, simply because it is not practical. They use self-selected samples and then study the data to find out how "similar" they would be to a random sample. (In fact, even random samples can be quite biased because people have a choice of not responding and it is usually a certain type of people who don't.) 3. People who know versus those who don't The key with surveys is to ask people who "know something" about the topic. Maybe not much, but something. Michigan Consumer Survey asks 500 randomly selected US residents about the US economy. They can take a random sample because everybody in this country (I mean US) knows "something" about their local economy - they experience it every day. They do not ask anybody from the other 95% of the world population. They intentionally select a subset of world population that is likely to know the most about the US economy. How would you identify and randomly select people who know something about stock XYZ? If you know a practical method, let me know and we will do it. If we made a random sample of US residents or even just investors, it would yield few who know "something" about this stock. Asking people who know "nothing" does not ad any value to a survey, just cost. The cost of a survey must be practical. All Internet surveys are self-selected, non-random samples. Let's ask a different question. How similar to the general population of people who have a position in stock XYZ or are considering a position in XYZ are Silicon Investor members or AllPredictive members? Similar in respect to their sentiment. So how can we identify a "population" of those who know something about stock XYZ? The presumption we make is that if you come to our site and make a prediction for XYZ stock, you are interested in it, you have done at least some research - you know something about it and your knowledge is not trivial. 3. Manipulation The discussion boards are full of it, when people do not call each other names. (So far, this one has been to the point and I hope it will stay that way.) Then a little bit of an info that seems to be truthful, but still not confirmed. With your reasoning, nobody should bother reading them. I am not going to argue with you whether people should or should not. They do and many, even pros, do it not for an entertainment but for "ideas". You sound like if a company would hand pick members to pitch their company and its stock at AllPredictive. I think company executives would be very careful about something like this. A group of pump-and-dump traders? It could happen. We have some procedures to lower a chance of manipulation and we will do more but nothing is 100%. 4. Conclusion As with any new product or service, some people are interested early and others are not. Marketers call people "innovators", "early adopters", "early majority" and "late majority" based on when they get involved with a new product / idea. I do not think that any amount of justification here short of hard correlation data would impress you, and maybe not even that – any argument can be countered. Our backgrounds, experiences, and personalities predispose us towards certain products or ideas and not others no matter how persuasive the arguments. I don't think I can convert you Ed but at least this discussion gave the "quiet majority" on this board an opportunity to listen. There is plenty that can be found on the Internet and libraries about statistics and random and self-selected samples. Please read if you like and form your own opinions. Ed, keep posting and if you can also provide some ideas how to improve this imperfect service, we would love to hear it. I promise the next time I will be shorter. Best,