Chapter 14 Thoughts on the stock market.
“If stock market experts were so expert, they would be buying stock, not selling advice.” Norman Augustine
I have dabbled in the U.S stock market for many years. I have had my share of losses and gains over the years. One of the conclusions that I have come to is that it is very foolhardy to try and make any short term predictions in the market. It is a sure path to ruin. Even long term predictions for individual stocks are very difficult. My systematic investments are in mutual funds. Index funds will work over the long run as well.
So the mindless and easy way to be wealthy in the United States (and probably wherever else there is scope to do so) is to start young and just automatically invest $200 or so into some index funds or mutual funds. I’m advising my son to do this. You don’t want to focus your life on earning money – on the other hand, not having any is definitely a nuisance and you want to be comfortable.
One of the greatest inventions is the ability to invest automatically by deducting from your bank account. It becomes really easy to do and you don’t even think about where the money is going.
Despite all these disciplined measures, there is a real allure to the stock market. Perhaps it is the hope that you can hit it really big. It appeals to the gamblers instinct that everyone has. I still look around for stocks that are so beaten down that there’s great growth potential. It’s the only leverage that ordinary investors like me have. The only other good leveraged investment, real estate, is in a precarious state at the time of this writing (December , 2010).
Anyway, I’m not trying to sell you on stocks, mutual funds or real estate. What I have noticed is that over the years, it seems to me the broader stock market does have a pattern. It’s hard to tell, and it’s not a completely intuitive cycle. If you expand the Dow Jones out to the maximum chart capability it looks like it does have a pattern of sorts. You can find this type of chart for the Dow Jones on any finance web site nowadays.
Looking at charts like this, I could swear the Dow follows some sort of a polynomial fit. The curve doesn’t look so irregular to me that you couldn’t possibly fit it to some sort of polynomial. It could be my imagination, but it is easy enough to test I believe. All you need is a supercomputer, and some data points.
We can pick each days Dow value as a data point for the past 40 years. This would give you 14600 data points. You can now try a fit to a polynomial to the nth power where n is 14600. You want to alternate pluses and minuses for the various powers to introduce some variation. Just looking at the curve, you can’t just rely on positive powers. This would give you and 14600x14600 matrixes to solve. I don’t know if supercomputers today can readily solve a matrix of this size. If they can’t we can just reduce the number of data points by using every other days value, for example. That would reduce it to a 7300 by 7300 matrix. You need to just tune the data points gathered to a manageable size for a solution by a supercomputer in a reasonable amount of time.
This is a brute force technique to see if you can get a predictable curve for the Dow Jones. The irony is that if it works, and it gets published widely, it will not work anymore because people’s buying habits would change due to the knowledge. So if you run this test and figure out the equation that works, your best bet is to tell no one and make your purchases on that basis. I wouldn’t mind if you shared it with me, of course.
You could run equations like this on an individual stock basis as well. It seems to me some stocks are just cyclical. Obviously, you have to pick the right type of stocks and a sufficiently long period (say a 20 year minimum) , for this to work If they fit a curve of some sort that extends over years, you could make a bundle by just buying at the right point in the curve. I know there are guys looking at technical indicators for stocks with curves and shapes (I have seen references to tea cup handles and other exotic shapes in technical stock analysis), but the polynomial calculation method to me seems like a better bet. It sure seems to the naked eye like a good fit ought to be possible. From years of looking at online transaction processing systems and fitting traffic patterns to polynomial equations, I have developed a good intuition for which curves are good candidates and which are not. It seems to me that the overall Dow Index and some individual cyclical stocks are good candidates to attempt a good curve fit.
The final point is that the curve fit doesn’t really need to be exact, although I have implied that with my brute force calculation methods. Even if you get an approximate fit that indicates general trends, you can make out like a bandit in the stock market. So if the fit turns out to be an approximate polynomial to the order of 4 but not an exact one, you should still be able to profit from the knowledge (after you do the right thing and share it with me for suggesting the direction to go in).
Discuss and enjoy!