Ed,
thank you for a very good discussion. Your comments are valid and we have considered each of your objections before going into this business. I have included some answers in your original message below.
>John Busby,
I might take this investment sentiment thing seriously if you had a panel of 50 prominent brokers who were utilizing their considerable knowledge and intuitions to make comment about a variety of companies.
A: The information you are asking for is available today. You can check analyst consensus, high, and low. Why are 12 month predictions so different from people who
- have access to the same data sources
- went to the same schools (Ivy League)
- work for the same kind of company (top Wall Street firms)
- use similar valuation methods?
They make it look scientific but it is not.
I checked the analysts' predictions for PALM today. The low is $18, high is $80, mean is $41. (Today price is $18.)
I do not know of any stock that is followed by 50 analysts. The number of brokers following Palm is 10. IBM is followed by 17 analysts. Most stocks have no analyst following, the rest have little following except for major stocks like IBM.
>Instead, this endeavor you are attempting seeks, for the most part, the opinions of lay investors who have not the experience nor expertise to validly assess financial statements, stock reports and press releases much less a sudden shift in market conditions.
A: I have seen the same argument made against consumer surveys in 50s and 60s. The naysayers have been proven wrong. Today, consumer sentiment survey is a leading economic indicator, tracked by Wall Street pros and US government.
The people who run Michigan Consumer Survey claim that, on average, consumers (not individually, but in aggregation) can more accurately predict national GDP or inflation rate than a consensus of top economists. Sounds counterintuitive? Is John Sixpack in Smallville studying national economy? (That is even though these economists, unlike analysts, have no motivation to fudge the numbers.)
>Furthermore, as noted elsewhere, you have a small community here at SI and therefore a very small sampling of opinions for each stock. Not only is this
unrepresentative, but it would be quite easy for a small
group of persons to skew the results for their own purposes.
A: I am not sure the community at SI is too small. Surveys of as few as 25 participants have been used in variety of fields. Anything over 100 is considered solid. Are there at least 1,000 people at SI? (Yes, there are.) If 1,000 people make predictions for 10 stocks from the top 100 stocks, you have 100 predictions per stock.
Regarding manipulation:
If there is an opportunity and incentive, you cannot avoid it. But it can be minimized. We are aware of the possibility and have implemented certain procedures to deal with it.
One procedure is "one member, one voice" principle. Our service uses only the last (the most recent) prediction for a particular security from a member. If you want to manipulate the prediction, you need to create multiple accounts which is against User Agreement and is a legally suspect activity, at least in the U.S. Another procedure is exclusion of the so called "statistical outliers" - numbers that are way out of the range.
There are other procedures and monitoring activities that we have in our arsenal but for obvious reasons do not want to publicize.
If you are concerned that some predictions are moving the price forecasts too far in one direction, you can look at what the median and 25/75 percentiles show. Is the prediction (mean) so high due to a few predictions or not? (Even that may be OK - when things are on a verge of changing, it is the few who notice it.) (This functionality is part of Premium service.)
>What is the value of a small sampling by non-professional investors really worth? Do you believe this would have any predictive value at all? I'm reminded of what investor sentiment must have been during the height of the technology boom. I bet the bulls predominated right up to the moment the bottom fell out of the market.
A: Do not forget, that the most bullish investors were the same analysts you prized earlier. Remember Henry Blodget? The herd mentality at Wall Street is well documented.
>You made the point about biases in press releases and financial statements. Fair enough, sometimes biases and downright dishonesty do creep into these things. On the other hand, my bet is the majority of companies with earnings make an earnest attempt an honest reporting. I don't see how you can compare these types of biases, an intentional or unintentional deception to ones stockholders, to the kind of biases people will express informally and anonymously over the internet.
A: Bear with us. We are just starting. What we have in the store for the future? We store members' predictions. Over the time, we can determine the accuracy of your predictions - overall, for an industry, and down to a particular stock. We will be then able to generate forecasts using data only from those who proved they know what they are talking about.
How would a manipulator go about it? S/he would have to create multiple accounts, be a good forecaster for a while (which is not easy), then suddenly skew the predictions. Then start again with a clean sheet. Possible? Yes. Easy? Not.
>Investors with some savvy will ignore your investor sentiment reports. Unfortunately, some lost investors will take the report seriously. They'd be much better off watching Louis Rukeyser once a week.
A: Investors should never trade based on just one type of data. Our position is that sentiment data is useful. Will the sentiment be correlated positively or negatively with actual prices? Will the correlation vary among industries or even individual stocks? We will find out as we go. Until then, it is responsibility of each one to make his/her own judgment of the data that are presented on our site.
Comparing to some people you mentioned, we never say "do as we tell you because we are smarter than anybody else".
By the way, we have had an interest from pros in lay investor sentiment data.
I would like to ask everybody who was interested enough to read up to this point to email us more questions (here or at feedback@allpredictive.com). I will put together some Q&A to address these and other questions you may have.
Thank you.
Peter Braun, Co-Founder, AllPredictive |
