Making Money out of Statistical Noise

Valentine's Day, 2003


Two swans form the shape of a heart with their necks as they float on Lake Geneva, Switzerland, Monday, Feb. 10, 2003, heralding the upcoming Valentine's Day on Friday, Feb 14. (AP Photo/Keystone, Martial Trezzini)

This page is inspired by something stupid that a certain market commentator said when the market made some significant declines. In January 2001, after the market had made significant declines, he maintained that buy and hold was not dead. In fact the fluctuations of the market proved only that there was a lot of variation, but still that on average stocks went up. The standard deviation of the market was high, but buy and hold still worked. I wrote him a short E-mail saying that this was total nonsense, and he said that the fluctuations in the market were just statistical noise. Of course, he never defined statistical noise in any mathematical sense. It was just arm-waving.

Oddly enough, even after the market has continued to go down, with few brief rallies, there are people still saying you should buy now. The market goes up over time. Maybe, but in the meanwhile, you sure can lose a lot of money. Does anybody believe that the people who bought Nasdaq stocks in March of 2000 will ever get their money back? Some of those stocks are not ever still around. Take WorldCom, for instance. Is it going to make a comeback?

Well, I will not try to define statistical noise either. While I do have a degree in mathematics, I prefer a common sense approach to the matter. Consider the chart below: Do the fluctuations look like noise to you? If so, I have some land in Florida that will be above water just any day. Buy now, while the land is still cheap.



Data provided by www.fasttrack.net.

The chart is the Nasdaq normalized to a value of 1.00 on the left. As you can see, it ends at about .75 on the right. This is a 5-year chart ending Feb 13, 2003. So over 5 years, buy and hold gets you a loss of about 25%. Is this what you want? For the S&P 500, even with dividends re-invested, you would be down about 14%. This is hardly a case for buy and hold.

Of course, there are 5-year periods when buy and hold looks pretty good. But it never really made any sense.

While someone who knows nothing about mathematics might be taken in by a commentator like that one, I think that nobody who looks at the chart will see noise. Instead one can see opportunities to make money on both the long and the short side of the market. There are trends, reversals, and more trends. Buy and hold nets you a loss over the last 5 years. Is that what you really want?

Or would you rather have timed the Nasdaq with Rutvolx?


Data provided by www.fasttrack.net.

Again, the chart is normalized to 1.0 on the left, but it ends as about 5.7 on the right. This is timing the long trades only. Going short would also sometimes have been profitable.

Now I must confess that I did not take advantage of all of the opportunities available to make money since March 2000. I was reluctant to short, and I needed to refine timing algorithms in order to be successful. I did not make anywhere near as much money as I should have. On the other hand, I am far from showing a loss over the last 5 years – that's for sure.

Technical analysis is the key to market timing and to making money from the trends of the market. Technical analysis, after all, is based on facts: namely the price movements. The technical indicators we use are based on what the market does. They are not based on hope or opinion. They are not based on some economic analysis that tells what the market “ought” to do. Sometimes there are false starts and failures, but the technical indicators allow us to cut out losses. The following chart shows the Nasdaq timed with Rutvolx. There are some gains and some losses. However, this is certainly an improvement on buy and hold.

Timing and trading systems should be trend-based and event-driven:

Use technical analysis to determine the trend. There are good market timing systems which the Fasttrack community have developed over the years. Fundbuster has a number of excellent signals.

Pay attention to events. I do not mean “current events”. I mean market events. When the trend changes, take advantage of it. Look at the market every day, run your Trade programs every day, check out the signals, and above all pay attention. Do not put on a position and go on a cruise around the world. You might not have anything when you get back. Do not, like the Motley Fool Mechanical Hibernation systems, decide to take action only on the first Monday or the quarter, or some such.

You have doubtless heard people say that you cannot time the market. This is certainly a difficult thing to do, and nobody can time it perfectly on a day-to-day basis. We have examples of timing systems that have not worked.

In general, systems based on market valuation do not work:

Even those who say that they do work admit that it may take years for “overvalued” stocks to become “fairly valued”, or for “undervalued” stocks to appreciate.

The most elaborate valuation model was the one developed by Elaine Garzarelli. She divided the S&P 500 into industry groups. Then she determined how these groups had been valued historically, based on P/E and projected earnings. If the group was undervalued, you were supposed to buy the most undervalued in the group. If too many of the groups were overvalued, you were supposed to sell everything. She had some success in 1987, but she sold at the bottom in 1996, and lost credibility.

While it may be the case that valuation matters in the long run, it is not a good key to timing the market. Besides, valuation models are based on what the market “should” be doing, and the market is not bound by such high imperatives.

Focus on the following:

Making Money. You are not in this to buy good stocks or mutual funds. You are in this to make money. There are no good stocks or mutual funds. There are only those that are going up and those that are going down.

Focus on technical analysis, not talk. Anybody can say anything, including a stupid remark on statistical noise. Technical analysis is based on the fact of what the market is doing. It is not based on what somebody thinks the market should do.

Ignore the charts at your peril. I have a page on that.

Remember: You can go short, and it is not Un-American to do so.

Again: Focus on making money.

Backtesting and My Tattoo:

Finally, I want to talk about backtesting and my tattoo. The ancient Greeks, and before them all the peoples of the ancient world, except, perhaps for the Jews, believed that history was cyclical. Spring, Summer, Winter, Fall. The rise and fall of great empires and the empires that replaced them. What has happened before will happen again. Neitzsche had a more complicated idea: in the infinite reaches of time, every event would be repeated.

The snake chasing its tail was a symbol of the eternal recurrence of this cyclical history.

In cyclical history, backtesting a strategy over many many years makes sense. Nothing fundamental - in fact, nothing incidental, has changed. So you just have to find out the technical indicators or whatever that worked the best over the last 100 years, and you have it made.

History, however, is not cyclical. Things do change. It is not just different this time, it is different every time. The fundamental characteristics of human nature have not changed, but the opportunities for good and bad have been immensely multiplied in our current time. The market trades differently at different times. Ken Huck, one of the programmers of FastBreak, remarked at the Florida Clearwater meeting (Feb 7 – 9, 2003) that whereas in the bull market setting stops only, in general, hurt performance, in the current bear market, the use of stops increases performance.

So, to use the symbol of the snake chasing its tail, the snake chases its tail still, but it cannot ever get to its tail. Its own body - the history itself - keeps history from repeating itself exactly. That is illustrated in my tattoo.

In the market, this means that a system that has worked very well in the recent past - say 1-3 years - is probably the best choice. It also means that the system cannot be static. Whenever you get a “sell” signal, it is time to start revising your system, waiting for the next buy. Harry Larsen of Fundbuster is continually improving his systems, to adapt them to the changing environment. That makes sense.

Finally

The market may surprise you with something unexpected, but as in the picture below, maintain your dignity.

[ Top ] [Introduction] [The Dogs] [End of Funds?] [Couch Doggie Portfolio] [Pick Good Funds] [Stock Trading Systems] [ Trading Systems ] [Simple and Profitable] [Journal] [ Timing ] [Selection ] [ Common Sense ] [ Statistical Noise ] [ Mechanical Hibernation ] [Best Ever ] [ LTBH Results ] [ Beat Market ] [ Technical Voodoo? ] [ Charts Do Not Lie ] [ Bad Advice ] [ Analysts ] [ Gamblers, etc. ] [ Relax ] [Links ]