A shibboleth on Wall Street and life is that we underestimate change.  I've heard it a million times in popular and academic papers. Zarnowitz adduced it in his study of forecasts 50 years ago also. I felt a fast study might be of interest. Do the big changes following big announcements in the early periods tend to continue or reverse. I tested it for 100 of the biggest changes. As usual the popular view is totally off. There is a big tendency for the big rises to be followed by big declines to an inordinate extent and the same reversal tendency for big declines. The study would have to be expanded to be of merit but it's a much better way of quantifying the effect than the subjective studies festering about.

UPDATE: I found 600 article (items) with the tag "underestimating change stock" on google and many of them are very interesting including an article on underestimating earnings announcement price movements and buying straddles to profit, also analysts not taking account of price changes before earnings announcements in making their predictions. But I didn't come across any that examined a large sample with a definite and non-overlappng data set like mine. My study shows that if you take all the important announcements, and look at the change in the first 10 minutes that are big, there is a significant tendency to reversal. I also looked at all the big 10 minute changes around 830 without regard to the announcement and found the same effect. I can say that at least at the microscopic level, with moves of about 0.2% expected, there is a substantial tendency to overestimate the impact of announcements. 

Adam Grimes writes: 

Thank you for that study and the perspective. It makes a lot of sense, and makes me ask a few related questions:

Something I've been wondering about is the claim that markets switch regimes faster now and that markets basically don't trend as well as they used to.

Two thoughts: 1) it's used as justification for the "death" of simple directional strategies… there does seem to be some evidence that we don't have the long trends of the 1980's that gave CTA-style trend following legendary returns for a while, but question 2): why do we assume this is linear? The people who discuss this would have us believe that we look back to the past and see markets that trend and have now fallen into a chaos (perhaps that's overly dramatic) where markets essentially no longer trend. Isn't it also possible this is cyclical, and that we could see more decades of those long, relatively "easy" trends in the future? The assumption is always that it has been driven by electronic trading, more competition, etc… but I wonder. (Always hard to truly understand the drivers… I guess understanding the effect would be enough.)

Not sure how to look at this idea in an objective way. Does this raise any thoughts?

Adam Grimes CIO, Waverly Advisors, LLC

My book: The Art and Science of Technical Analysis

Shane James writes: 

Hi Vic,

I concur with this.

Of note in many of the macro markets is that there is quite an imbalance in the price formation process.

In other words, within the idea of 'conditional heteroskedaticity' or 'volatility clustering' the relatively large moves do tend to occur contemporaneously and the relatively smaller moves also do the same,

The 'imbalance' I refer to above is that- overwhelmingly- the clustering of small moves proves the point more so than the clustering of large moves.

In terms of sign the smaller moves have more persistence than large.

Rocky Humbert writes: 

The money quote in the Forbes article that you cite is, "Find the trend, but don't sweat the details or the timing because you'll always be wrong." There are 4 sub-statements in this sentence. Which of the 4 are you challenging? And bear in mind that your answer must be consistent with your faith in the "Triumph of the Optimists".


WordPress database error: [Table './dailyspeculations_com_@002d_dailywordpress/wp_comments' is marked as crashed and last (automatic?) repair failed]
SELECT * FROM wp_comments WHERE comment_post_ID = '10748' AND comment_approved = '1' ORDER BY comment_date




Speak your mind


Resources & Links