Finance Wanted.

By Vadim Rob

The truth is out. The traditional argument as to whether technical analysis has any legitimacy as compared with traditional buy and hold stock market investing techniques lives on and the few of us as money managers who seem to have survived, smile knowing much of the argument is amiss in what it is really asking.

I have managed to survive and do well as a money manager. I have watched market advisory services come and go over the years with new ones coming forward with new claims and new participants entering the market ready to eat it all up (or be eaten).

But, what is it that has enabled me to survive as an active manager? Is it my phenomenal skills? Is it this magic thing called "Technical Analysis."? Or is it just simply being in the right place at the right time?

Certainly two of three of those choices could be random. But one thing is not, and that is Technical Analysis. In the traditional sense, as one might read in books, I have never ever found any technical analysis tool to be of any benefit to me in the development of a trading strategy (barring moving averages which I use to introduce intentional delays or to create zones to trade from).

So if I could never use them, then what exactly is it that perpetuates the myth (at least as I see it)? The best guess I can come up with is failure to do a simple test of the obvious. The survivorship bias of most people not staying in the game for any length of time results in new participants going back over the same nonsensical stuff again and again.

So, I decided to conduct a test (actually, I did this 15 years ago but will duplicate it here for you). Real science (as I have done for years) and resolve, once and for all (or at least in portion) this debate.

I conducted a test on the efficacy of stochastics (oscillators in general) as stand alone trading devices. The text book method for the use of these tools in trading is that a condition below 20 means the market is "over-sold" and should be bought. If the oscillator goes over 80, it is then called "overbought", and should be sold.

This all seems very scientific. And, if you look at a chart, your human eye will gladly pick out all the highs and lows the oscillator picked (to the exclusion of all else). Convincing! But as I said, this is science, so we won't rely on the human eye.

Oh yes, you might argue. Stochastics (oscillators in general) can be used in other ways than for overbought or oversold conditions. I full heatedly agree. But the principle in the following tests will still stand. So, bear with me...

Let's look at a scatter plot of a stochastic oscillator for various levels five units of time into the future. This picture will tell us everything we need to know about stochastics (on the given time frame) without any messing around (this is what scientists do) -

The graph is made up of over 3000 data points on a 5 minute chart, looking 25 minutes into the future. Now before your brain shuts off, don't worry. It is not that scary. Simply, it shows us that on this particular test, for values of the stochastic oscillator above 80 (that is supposed to be a sell signal) the returns vary from -10 to +10 points overall. In fact, the returns vary from about -10 points to +10 points overall. Over on the left, it does show there were about 5 events (grey dots) where there were declines of as much as 20 points, but over all, the graph is random (evenly distributed). On the bottom of the graph, we can see, below 20 (that is supposed to be a buy signal), the returns are also distributed from about -10 to +10 points. The middle ranges are also similar. This tells us that the distribution of prices 25 minutes into the future on a stochastic oscillator are... (oh no Rob, don't say it) RANDOM!

There, I said it. Random!

If you had traded using the 80/20 rule over the period of this test, holding 25 minutes on each trade, you would have lost over $3000 not including commissions. Next time you think about using an oscillator in the textbook traditional sense, remember this.

The fact is, this test result will hold true for just about any time frame on oscillators and just about any other technical analysis indicator. As the time frame increases, it may become more reliable (at least on stock indexes). In many other cases and markets will actually be the opposite (ie. Buy 80 and sell 20).

Maybe the age old debate will linger on? Maybe I have put it to rest? I think not. Though I am sure I will get some responses to this (and I hope I do) so you can send me your technical indicator. I will gladly test it for you and send you the scatter plot with an interpretation (subject to my own scheduling). Fact is, I have never seen a technical analysis indicator hold up to this type of scrutiny and this is one of my "nice" tests ;-)

This article is not to discourage the use of oscillators or other technical analysis tools. Quite the contrary. Tools do have their places depending on the intent and design. I always encourage trading based on a solid premise. Getting at the testing of the premise is key, and scatter plots is one great way to cut through a lot of garbage and find truth quickly and save you from a lot of heartache. It only took a couple minutes to set up this test. I encourage you to do the same, or, send your indicator or trading system to me and I will test it for you and help you with logic and improvements if I am able...

When you find a solid premise to work on that holds up to stringent testing, you can start making some good money trading with it. Knowing where you are through testing makes the psychological component of trading easier, lending to the mental success cycle required to succeed financially. Asking the right questions can put debates to rest and lead you to greener pastures.

For more info please visit my blog at http://eminiforecaster.com/blog