## Friday, September 24, 2010

### Market timing futility

kw: observations, investing

Look at this image for a while and let it sink in. It shows the correlation between the market change of AT&T stock on one day with that on the day following.

It was produced thus: At Yahoo Finance I entered the symbol T to get a stock quote, then in the navigator on the left I chose "Historical Prices" (the 4th item). In the page that comes up I set it for daily and used the default time span (from some time in 1984 to present). At the bottom of the page is a button for downloading an Excel worksheet; I clicked that.

In Excel, I copied the "Close" column (NOT "Adjusted Close") to a new sheet and labeled the column "Day1". I copied it again, starting at the second value, to the next column and labeled that column "Day2". This sets up the basic correlation. I highlighted the two columns and made an X-Y chart, then changed the symbol color to dark red. I set the scale on X to -3 to +3 and that on Y to -2 to +2, so the plot is approximately square. There are several data points in the range of 30 to 70, where the stock split. This scaling gets rid of those from the view, and focuses on the day-to-day variations in the range of a few percent or less, which is 99% of all the data.

Just for fun I performed a regression using Excel's Data Analysis package; the correlation coefficient was 0.0003, which is even closer to zero than I expected. Consider this visual proof that no matter what the stock did today, it has no bearing on what it will do tomorrow. Any slice of these data in any direction whatever is a random distribution, centered on zero, with heavy tails, meaning that larger jumps are more common than for a Gaussian distribution.

The only "market timing" technique that works is to hold a stock for a very stable company for a long enough time that the general rise in prices will bring you a gain, then perhaps watch it for a day-to-day variation that will snag you an extra percent or so, then sell. Of course, at that point, in what will you invest now? All the other stocks in the market have been going up also. So the basic method of "buy when you can, sell only when you need to spend the money" actually outperforms any "technical" scheme.