It is every stock-market chartist’s dream to discover the secret of always making money! If stock prices indedd are periodic (ergodic), then making megabucks shouldbe as easy as pie…just program your computer to recognize the bvalleys /low spots in the cycles, and buy. Then, when the cycle turns up, you simply have the computer send out “sell” orders.
The stock market chartisit claim such a straegy is possible…yet, if this were so, wouldnot the market have so many attuned investors, that this periodic behavior would soon disappear?
I’ve been looking at stocks for years now, and while there are certainly ups and downs, it is hard to discern any recognizable periodicity in their behavior.
Anybody know if fourier analysis has been applied to stock pricing?
And, how successful have the “chartisits” been in predicting stock prices?
If you’ll forgive the lack of cites, I can offer a little info on the subject. Depending on how you extract the “cycle factor” from any time series, you will find the appearance of irregular cycles in almost any data.
I did a lot of that in forecasting analysis where I worked. Winters’ Model and other such things that extract trend, seasonality and other theorized components. At one point I was convinced the 4-year pattern was pretty solid. Even tried to persuade the powers that were that it was reasonable to predict just the cycle factor and let the model take care of reconstructing the forward curve.
But as I learned to my chagrin, there’s a fundamental law (much like Murphy’s) that applies to all this.
You can track a series for years and observe all its regularities and predictablities and have yourself convinced you know precisely what will happen next. As long as you continue to observe, things remain solid and reliable.
BUT, at the instant you do make a prediction, things will change – in an opposite direction from your prediction.
Another thing: get yourself a reliable random number generator (a pair of dice will work for lack of better) and collect 1,000 data points. Construct the model for trend and seasonality and extract them. You will find a “cyclical factor” remaining. This should hellp alleviate your conviction that cycles really can have value.
Good luck!
My understanding is that “chartists” generally do no better than if they’d guessed randomly.
However, any chartists out there who are making money are probably smart enough to keep their techniques to themselves. As you point out, any publicly known regularity would quickly disappear. So it’s conceivable that there are technical trading techniques out there that work but are secret.
It is surprisingly widely received that the Stock Market follows a seven year business cycle. It has even been suggested that this is linked to a seven-year cycle in sun-spot activity(!) (This supposed “link”, apparently refuted by Keynes around the Great Depression, still pops up from time to time.)
Sun-spots or not, it seems you have to be somewhat flexible with your notion of “seven-year-cycle”.