From theory to practice. Part 2 - page 98
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Thank you again! Now that's constructive dialogue in this thread. I have a lot of creative thoughts, but no knowledge of how to approach the issue in the right way.
Zhenya, I like it. Make a simple EA, for example, when registering a wave up we buy (or sell) and down we flip. Then you find in the optimizer the value of the minimum step at which you get the most profit for the whole period.
But then the following picture is obtained: in some years the profit is stagnating, in other years it is falling, in other years it is growing. It turns out that we have picked the median movement size for the whole period, but within it there are periods with another best zig-zag wavelength size. That is, the median length changes all the time, but its change is not abrupt. It can hold for years. The question is how to find a tool that will break the period into segments with the same price behaviour.
You should probably look at entropy rather than distribution density. If you're looking for predictability.
May make sense when looking for dependence (on the previous knee, for example) and considering the entropy of the joint length distribution of the two subsequent knees. The correlation or deviation of the copula from the theoretical one for SB can also be useful. Entropy for a one-dimensional continuous distribution is just some mathematical dabbling, in my opinion. There is practical value in the joint entropy (information) of discrete distributions.
Thank you again! Now that's constructive dialogue in this thread. I have a lot of creative thoughts, but no knowledge on how to approach the issue in the right way.
Please, I'll share what I can.
If we talk about the median of knees, it is also useful to compare it with the theoretical value for SB, which is z0*(1+ln(2))
The interquartile range and its ratio to the median and comparing it to the theoretical value for the SB may also be of interest.
Eugene, you have managed to get a lot of forum participants interested. Use it, they will together help you to approach the question. But... only in case your pictures can be trusted. It means that they will be verifiable. That is, anyone else will be able to repeat the calculation and the construction, in particular the histograms. Then it would be possible to find out if the results you obtained are reproducible. The key word is "reproducibility". For this you must always know exactly: the source of the data; the range covered (the sample); and so on and so forth - up to a set sufficient to repeat your work on the data given, but without your involvement. In scientific articles, such information is often collected together in the "description of the experiment" section. It is to check reproducibility. Without reproducibility of results, there is no point in evidence-based work.
I don't understand what you're getting at. Repeat what you want, who wants what I'm getting at?
I don't understand why you wrote all that. Whoever wants it and what they want, what am I interrupting?
It is the lack of specificity that prevents. How can I check, for example, this "I've got the following histogram for EURUSD with ZZ step 100 (5 pips)"? How can we hope to reproduce it?
Where and what data did you take; for what period; what was the chosen zigzag; what parameters did it have, except for the pitch? The absence of this information is the obstacle you (I think, involuntarily) put before those who would like to repeat your calculations.
The lack of specificity is an obstacle. How to check, for example, this "For EURUSD with a ZZ step of 100 pips (5 digits) I got the following histogram:". How can we hope to replicate it?
Where and what data did you take; for what period; what was the chosen zigzag; what parameters did it have, except for the pitch? The absence of this information is the obstacle that you (I think unwittingly) put before those who would want to repeat your calculations.
Did you see somewhere in there the inscription - 'repeat such a picture after me' ?
OK:
ZigZag with a single parameter - Step = 100 pips by five digits.
Symbol - EURUSD
For the period - look at the last date in the file, I did not download any further history.
Do you want to repeat it? Or you just want to talk?
Another sucker.
Two regularities are explained to you in Russian on the series of SB increments.
1. The set of sums of these increments always forms a normal distribution
2. SB tends to move away from the starting point. And if, on average, realisations give MO = initial reference point, then a particular realisation always gives an earning opportunity.
I remember this Wizard. He is a faithful friend and associate of the undercutting Aliosha. But, that doesn't stop him from being a sucker.
Why be rude? I knew a quantum Alexey Bondarenko, who used to work at a couple of Moscow mini hedge funds, and then he went to the U.S. and kind of went to jail there. I communicated with him maybe 3 times, and then long ago, when I was just starting out. He seemed smart at the time, but by today's standards it is a child's play.
Let'sdeal with SBin a civilized manner .
I will immediatelynote among amateur enthusiasts, the theme of making money on the SB is fairly common, as in fact some people do not let go and the idea of circumventing the law of conservation of energy, creating a perpetual motion machine, it is normal, even people with mangelingale at first dabble, and then strongly involved, contrary to common sense, it is a classic trader's deviation. By the way, read Newton's biography, he was an alchemist to the end of his life, he liked it , and that's all , at the stock market he alsospeculated on emotions , intelligence and delusionsoften go together.
So you say:
1. The set of sums of such increases always forms a normal distribution.
If there are no overlapping sets , evidently yes. But what does it give us? It is not the gradients we trade, we trade the sum, and for the gradients the ME (total, windows it does not matter) converges, but not for the sum, it is unpredictable as well as the initial (aggregate) series. You can't predict any sum of the following increments, so you can't trend trade. You cannot predict the MO in any window, so you cannot trade reverse. There is no basis for "high-level" statistics, from e.g. seasonality of autocorrelations to something more ornate (all kinds of "market conditions", etc.) that can catch MO . All interesting statistics are non-stationary.
2. SB tends to move away from the starting point. And while, on average, realizations give MO = initial reference point, a particular realization always gives an earning opportunity.
It's not quite clear what you mean here, are you talking about incremental\returns or aggregates (price) and what "specific realisation" means. We need to know what can be predicted and what can be traded. You better give us an example. It is clear that some piece of sample, retrospectively, may contain phantom patterns, be trending or flat, seasonal, etc. What use is that to us? We know that we cannot predict what will happen next, and this is the main thing. : )