From theory to practice - page 1509
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
It is unacceptable to publish personal correspondence explicitly without the correspondent's permission
at best, you may only refer to an unknown source in your own words
I couldn't resist... He will forgive... He hasn't been on the forum for a long time - he's probably lying in a rubbish bin. He's suffering from hunger and cold... We should help him. How? I'm as naked as a falcon, I can hardly walk to the factory...
When I look at your perfectly symmetrical rows (I don't know how you do it, but maybe it doesn't matter anymore) I am reminded of Doc's words from a personal correspondence:
Because the distribution of price gains is symmetric, and this symmetric distribution is preserved in the sliding window.
Something like this is what I need to achieve on M15.
2018.04.17 00:57
The yellow line is a moving average with a window of 100. Almost perfectly flat.
I think back (may he forgive me for posting these screenshots) and weep bitterly - how many suffering, intelligent and talented people the market has scattered in different directions... Can't even count...
Or maybe Doc, on the contrary, found what he was looking for and just no longer wants to communicate with us pygmies? It's better this way.
Decrease the point size of the price movement with increasing volatility, no? Filtration does the same thing as yours, only when volatility increases the ticks are not added at all (or very rarely)?
Reduce the point size of the price move as volatility increases, no? Filtering does the same thing as yours, only when volatility increases the ticks are not added at all (or very rarely)?
This whole topic is about this.
When we talk about incremental probability density, the first thing we pay attention to is the heavy tails, the outliers. But, in terms of MO, that's not the main thing - what's important is that there is no asymmetry. The distribution must be strictly symmetrical at any significant sample size.
The doc took my data on AUDCHF (I'm afraid it's just a lucky sample), where returnees at any time form a symmetric distribution, trained the network on it and it started to produce 100% accurate (or nearly so) forecasts. However, the data was ticks, though thinned... He was hampered by spread and commission... He decided to achieve the same symmetry on OPEN/CLOSE M15.
Alas, this is where the story ends - his traces were lost. We do not know whether he has solved the problem - the transformation of the initial asymmetric series of increments to symmetric form. But Koldun, on the other hand, seems to have solved this problem.
This whole topic is about this.
When we talk about incremental probability density, the first thing we pay attention to is the heavy tails, the outliers. But, in terms of MO, this is not the main thing - what is important is that there is no asymmetry. The distribution must be strictly symmetrical at any significant sample size.
Doc took my data on AUDCHF (I'm afraid it's just a lucky sample), where returnees at any time form a symmetric distribution, trained the network on it and it started to produce 100% accurate (or nearly so) forecasts. However, the data was ticks, though thinned... He was hampered by spread and commission... He decided to achieve the same symmetry on OPEN/CLOSE M15.
Alas, this is where the story ends - his traces were lost. We do not know whether he has solved the problem - the transformation of the initial asymmetric series of increments to symmetric form. But Koldun, on the other hand, seems to have solved this problem.
It's not a problem for MO to simply do the balancing of classes. The problem is quite different - the lack of regularity in the increments (read - cycles). This could have been put a fat point on it like a year or more ago. But they, instead of admitting it - either silently merge, or throw meaningless pictures around.
Point 2 especially guarantees futility, market participants destroy repetitive patterns that could be exploited
From a global point of view - all is futility, vanity and languishing) From a more down-to-earth point of view - studying through the matstat exactly how any patterns are destroyed (how non-stationarity is arranged) can be quite useful.
This is not a problem for MoD - just do the balancing of classes. The problem is quite different - the lack of regularity in increments (read: cycles). This could have been put to a fat stop like a year or more ago. But they, instead of admitting it - either silently merge or throw meaningless pictures around.
Maybe so... I don't know... Of course, one can only believe the state. On the other hand, a person with a solved problem will never show the state. Contradiction. You don't know who to trust.
Maybe so... I don't know... Of course, you can only trust the state. On the other hand, a person with a solved problem will never show the state. Contradiction. One doesn't know who to believe.
This overwhelming belief in conspiracy theories...
This overwhelming belief in conspiracy theory...
I can only say for myself that in case of constant symmetry of probability density, without outliers, the increments form a completely stationary random sequence. And such series, according to Kolmogorov, can be predicted.
I can only say for myself that in case of constant symmetry of probability density, without outliers, the increments form a completely stationary random sequence. And such series, according to Kolmogorov, can be predicted.
Such series are called residuals (when they do not contain any useful information). They themselves are of course predictable, but they do not predict the original series.