Building a trading system using digital low-pass filters - page 14

 
bstone:
And now let's call the "elliotniks" from the next branch and you can be sure they will foam at the mouth and prove that we are facing a classic 5-wave, followed by a developing compound correction X-Y-Z.


What is it then?
 
Mathemat:

The stage at which I naively thought that returns are distributed according to the normal law has long been passed, thanks to Rosh. And Prival recently, a page or two ago, posted a picture showing that normal distribution doesn't rule here. There is more complicated mathematics here.


There's not just a picture, there's a rigorous mathematical proof.
 
Mathemat, enough with the letter differences. I showed a simple example, using a normal distribution. If you're not happy with the normal distribution, then go ahead and transform it to what you want. There are methods available. The volume of data allows you to do this with sufficient accuracy.
 
Integer:
bstone:
Now let's call in the "elliotics" from the next thread and rest assured they will be foaming at the mouth to prove to you that we are facing a classic 5-wave, followed by a developing compound X-Y-Z correction.


What is it then?
Now that is a very good question. If you have an idea of Elliott Wave Law, then you should know that Elliott was based on market psychology (i.e. stages of confidence, doubt, fear of investors). But here's an interesting question - where in the dumb normal distribution of random numbers did human psychology with all its subtleties come from? :)
 

Roman, OK, I'll show you that this is not just an excuse, but a real difference.

What is sigma equal in your generation? Now take it and see how many values in your series differ from the centre (zero) by more than five (!!!) sigmas modulo. How many? According to the Gaussian law, 10000 * 0.0000006 < 0.01. I.e. the probability that you will encounter at least one such deviation is very small (they are the thin tails). At the same time, the real data will contain about 20 such samples among 10 thousand (I already checked).

 
Mathemat, you misunderstand me again. My position is that if I took a simple normal distribution (horrible, with thin tails, no anomalous sample terms, etc.) and got a visually very similar to the real price series, then if you take a sample of 10,000 elements obeying the real distribution with a given accuracy, the result is much more reliable (just as much as you need). Although we won't see any visual difference.

Make a histogram of the real distribution of desired accuracy and use it to artificially (numerically) generate the same distribution from a uniform one, as they do in the textbooks for normal. And then repeat my example as many times as you want to get the synthetics you need.

It should be pretty clear now.
 

bstone

I think the problem is that it is theoretically impossible to make money on the series you have generated

 
Prival:

bstone

I think the problem is that it is theoretically impossible to make money on the series you have generated


Don't look at the generated series, but at the "price series" that is obtained by integrating the generated one. Was there any money to be made on it? I think so :)
 
bstone:
Integer:
bstone:
Now let's call in the "elliotics" from the next thread and rest assured they will be foaming at the mouth to prove to you that we are facing a classic 5-wave, followed by a developing compound X-Y-Z correction.


what is it then?
Now that is a very good question. If you have an understanding of Elliot's Law of Waves, you should know that Elliot was drawing from market psychology (i.e. the stages of confidence, doubt, fear of investors). But here's an interesting question - where in the dumb normal distribution of random numbers did human psychology with all its subtleties come from? :)
It was an attempt to explain the pattern seen.
 
It was an attempt to explain the pattern seen. <br / translate="no">

Well it seems to me that this pattern was pulled by the ears. And then there was a fashion to find Elliott waves everywhere (football team performance charts, tides, etc.).

And I think Elliot was trying to see a pattern in the results of integrating a series of random numbers :)