thank you for your comments everybody, those that suggested that there is nothing to this, well I guess those would be happy if I laid down all the formulas and source code my friend so far developed once it is proven I would have the satisfaction of people saying "yeah that's nice" but sorry guys I am just not that dumb, I think very few on ET are that stupid PS I am a guy who uses 60 000 bit variable key none HD encryption, so go figure if i am the type that explains details to school children
The only bit variable figure anyone in ET cares/should care about is how much money you actually made trading. Doesn't prove skill or brains, but it is better to be lucky than smart. There are legions of smart but poor people in the world, and they are not very interesting from a trading perspective.
First, once people started charting, they looked at the chart and said, "Hey...there are trends here. Let's try a moving average." And they did, and they were profitable. And things were good. Then the moving averages didn't work as well anymore, since everyone was using them. There's too much chop they said, so let's filter the noise. And lo, channels were born. And they were profitable. And things were good, for a while. Then the bands didn't work any more and it was all chop and no trend. So the RSI was invented. And it was profitable, and things were good for a while. Then someone noticed that,"Hey, these charts look like waveforms. Let's try a fourier analysis." And they did. And it was profitable... for a while. Then someone said,"This stuff doesn't repeat that often. It must be based upon (insert mantra): 1) Chaos theory, 2)Fractals 3)Gaussian wave propagation" And they used it. And it was profitable... for a while. Then someone looked up and said, "Hey, all this stuff is way too complicated. Let's go back to the moving averages..." Curve fitting is curve fitting. It makes no difference if you extract another .003% profit with .05% decreased DD. The system when developed in that manner is bound to fail. Spoken with some experience on this matter.
If you watch the water running down the windshield too closely without watching where the CAR is going, you will crash into the car in front of you and the path of the water won't matter. Risk management is of utmost importance.
With only 16 posts, and a reply like that! Welcome my friend To the Original poster: Very interesting thinking. I did research on something simular, but of a single point in a - let's say windshield - that is less random. If you create a small crack in a windshield, over time the crack will grow, but never in a straight line. Sometimes from different points, or a curve. Would you apply it as a Algorithm? -Kastro
Hi I'm an experienced day trader (several years) and computers engineer (more than 20 years). I'm running in this forum and other sites, a set of articles about "auto trading" and one of my next up coming articles will deal with the exact issue, can we find a - so called - "predictable" behavior behind random chart ? Generally speaking the answer is "Yes but.." and there are many theories around it (I've developed and use a mathematical theory that is eventually implemented by a multiple computers array and it reaches a very good annual results). If you are really interested in diving into it you are more than welcomed to follow my articles. Have you heard about the "Eliot's Waves Theory" ? This is a very traditional theory that connects between fibonacci series and stock charts. Best Regards Trady1
I would not say that. If what your system is doing is curve fitting, it must fail eventually. But what people don't understand is that you don't want a system like that. The key to all systems is to recognize when the market dynamics that matches the systems dynamics are no longer working. If what your system is trading is based on real dynamics of what you are trading, then you will always know when that dynamic is no longer in play. If the dynamic is in play again, the system will start working again. The problem with curve fitting (and most systems developed in existance by amateur system builders) is that the human brain has no idea why the given (polynomial) estimation is working or has stopped working, and then you have to rely on more vague statistics to tell you when it is time to turn off the system. That is why people on ET will tell you that the best systems are the "simple" ones. They don't realize that simple doesn't mean simple in terms of being free of abstract mathematics, they mean simple in that they can verbalize it and instantly, with their own brain tell when it stops working. What is funny about this definition of "simple" is that these types of "systems" are infinetly more "complex" than the mathematical systems. Think about it, it takes the most complex object in the known universe to be able to come up and understand and monitor them. No random search by a computer would likely ever come up with them. This is the Penrose argument in "The Emperors New Mind." http://www.amazon.com/Emperors-New-...ef=sr_1_1/104-3364832-9951153?ie=UTF8&s=books nitro
To the orignal poster. This subject is near and dear to me. In fact it defines me. I cannot get into it, for I would spend my entire life posting on ET about it. So I stay away from posting about it. If I were going to write about this subject matter, I would do it on my own website as a blog. nitro