I’m preparing a lesson on recursion for a high school math class, and was looking around for stuff on Fibonacci numbers when I came across a whole stack of literature on Fibonacci analysis of stock charts. Now, this seems to me to have all the validity of biorhythm or astrology charts, but searching Google and Amazon failed to turn up any solid debunking. Can somebody help? Pointers to studies would be nice :).
Somewhat related are a couple questions about chaos and the market. As I understand it, a chaotic system is one in which a small change in initial conditions leads to a big change in final conditions. Does the market really fit this (or some other) definition? I would think that any small events would get absorbed by all of the other small events and hence be negligible. Am I wrong? And, supposing the market is chaotic, how would that make it any easier to predict where it’s going to go?
The problem is this can’t really be tested. You can never re-create the initial conditions so you can tweak them. You can do it in simulations, of course (which is how much of chaos theory got its start, with weather sims) but a simulation of something as complex as markets doesn’t take anything near the required number of variables into account.
Financial markets fit any reasonable definition of what is chaotic. Knowing markets are chaotic doesn’t make them easier to predict but it will at least stop you from wasting your time with linear analysis. Some people maintain that since markets are chaotic they are inherently impossible to predict. I suppose they haven’t heard of non-linear math.
From a chaos theory perspective, fibonacci numbers would be considered by their proponents to be strange attractors.
No one who uses fibonacci numbers would agree that they work in a vacuum so they become impossible to isolate as variables in any easy analysis.
The goal in such analyses is to find a previously undiscovered correlation and exploit it before anyone else hears about it. Once the correlation is known is becomes mainly unexploitable. (Not everybody can get rich.) A classic correlation is between stock and bond prices (in the inverse). So people throw a lot of different math at a lot of economic data and see if they can find something new. Fibonacci makes sense in that increases are generally based on recent, previous increases (or decreases for that matter). Probably a lousy fit but if no one had seen it before, ka-ching (for a little while). A lot of newly discovered “correlations” are just random effects instead and disappear during the next business cycle. It’s very chancy stuff involving big $.
I have seen a lot of want ads in the last year by New York firms looking for computer geeks with a good understanding of economics and statistics. Guess what those companies want to do.
8 years on the institutional sales desk. So called Technical analysis ranks up there with goat entrails and tea leaf reading. That goes for Fibonacci numbers as well. I had one colleague who was a Fib fan. Didn’t make her rich or even right very often.
Sorry, though, don’t know of a good debunking site. Analytic traders such as the O’Connor crew would laugh their ass off at any of these chartists and technical analysts. Geomancer would be a better description.
Most chartists and the like show how their model works in the past, and very few of them make real money in the future.
Past performance is not indicative of future performance.
PBS did a great video about some guys who did a math formula to predict the market that worked
& made a lot of money, that is, until the market changed in a way their forumla couldn’t predict.
“Mathematicians, traders and financiers have studied stock markets ever since they were invented several hundred years ago, but no one has yet managed to devise a foolproof system for working out what will happen next. But financial analysts can use historical data and mathematical calculations, such as moving averages, on balance volume and the stochastic oscillator, to try to predict future changes in the stock market.” http://www.science.org.au/nova/047/047key.htm