Do you think polling results can have any prediction power on stock prices? estockpoll.com

I do recall Ken Fisher writing something about the impact of election outcomes on stock returns in his "The only 3 questions that matter" (or something like that) book, so it doesn't seem like a stretch to extend that to polling results.

But do you think that peoples' opinions on stock prices can be used for prediction?

Why polling? How about just implied sentiment from futures and options? There is money behind that, not some random poll.

The rationale for these kinds of polls is that: 1) the voters tend to be outside the usual population of professional analysts who work for large financial firms that have incentives to say excessively nice things about the companies that might hire the analysts' firms for financial services; 2) some voters might have especially close natural connections to the analyzed company and have near-inside information; 3) the participants add diversity of opinion in using non-mainstream models. Of course, the downside is that some voters just vote their position and others are nutjobs with crazy magneto-astrology models.At the very least, it's another source of data from a distinct subpopulation of the market.

QuoteOriginally posted by: ChicagoGuyDo you think polling results can have any prediction power on stock prices?you know how you poll on stocks? you buy or sell them. that's how you poll.

sometimes we hear about overprice or in opposite. in what degree these judgement objective/ subjective/ quantified?

Why polling? How about just implied sentiment from futures and options? There is money behind that, not some random poll.True, you can imply market sentiment such as volatility from derivatives' prices. However, a model always has to be assumed to do this. Then you run into the problem of whether the underlying model that is being used is correct. Polling doesn't require an underlying model. sometimes we hear about overprice or in opposite. in what degree these judgement objective/ subjective/ quantified?I am guessing you can quantify polls as you would quantify any polling results with basic statistics.

I presume you mean opinion polls? I guess you need to look not only at the reliability of the poll but also any models that factor the outcome into your financial gain/loss. In the UK we often find that polls can be quite accurate - to within a few percent of the actual outcome - but that few percent is insufficient to know what government you're going to get (for example). Also there is adverse spin before an election - usually something like "If Labour wins the election then markets will tumble" which generally turns out to be quite untrue even though commentators also say that the possibility of a Labour win (in this case) is already factored into the market.I would be skeptical about any models which factor polls into predictive models, except perhaps in Greece just now!