Why is the normal distribution not normal? - page 23
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
Here is a graph of the distribution a(n)-a(n+1)
And this is the same TF, but a(n)-a(n+10)
Much better, isn't it?
I'm not talking about generation of new TFs but taking increments not with the previous bar but, for example, with the tenth one.
x1=a(0)-a(10); x2=a(1)-a(11); x3=a(2)-a(12)........
If I'm stupid, tell me. I will go away with shame.
muallch, the larger the timeframe or interval, the more the distribution of the first price differences approaches the normal one. this is a fractal characteristic of the price.
I'm not talking about generating new TFs, I'm talking about taking increments not with the previous bar, but with the tenth bar, for example.
x1=a(0)-a(10); x2=a(1)-a(11); x3=a(2)-a(12)........
If I'm stupid, tell me. I'll walk away in shame....
You are looking at the "overlap" case - you have a 10-sample wide window, but the windows themselves are overlapped by 9 samples. You can get very convoluted and smooth correlation relationships that have little to do with the real picture. Do you want to study this mess?
Yurixx wrote >>
to Neutron, Avals
...
Thank you, Yura. I see.
The bigger the timeframe or time interval, the more the distribution of the first price differences is close to normal. this is a fractal characteristic of price.
Rather, it is the law of statistics, according to which the sum of SV with arbitrary distribution tends to normal distribution in the limit and the noticed property of price series is a special case of manifestation of this law.
P.S. I with easy hand of Swinosaurus have become interested in looking at process of pricing on the Fodov Market. For a long time I wanted to compare conditions of work on FR and Forex. While I look (like an artist of course) at MMVB stocks quotations.
For example, here are Lukoil shares. At first glance, nothing unusual... But with almost zero correlation between one-minute bars their FRs look like the work of an impressionist:
Wow... >> I've never seen such needlework on Fora.
to Neutron, Avals
You guys should have looked at this advertised "advisor". It is such a child's handheld calculator that looks into the future and calculates how many pips you could earn if you trade on the zigzag peaks. Hence the "genius" idea that trading on the zigzag with spread+1 parameter is optimal.
Looking at this achievement, one can understand that getch understands optimality as an efficiency, i.e. the percentage that could be earned compared with the maximum that his calculator gives. Obviously, one cannot earn more than that in principle.
And you're talking about real EAs, for which the KPI is a few percent at best, and that's not how you've set the task. Obviously you have not read much fiction in your childhood. :-)))
You people are stubborn and no wonder. You are incapable of questioning and taking a somewhat different view of things that are stable for you.
For some reason, logic doesn't work. One only starts to listen when a bank statement with N-digit sums not in rubles is shown. Doubts and mistakes are not a sign of dementia, but rather the opposite.
Not bad mate training and skills are applied in vain. Many of the results shown here, which are academically unpromising, are successfully applied in trading. And it's not a matter of mathematical training, misunderstanding of terms, etc., but in the initial incorrect applicability and interpretation of the data.
I will not show screenshots of bank statements. I remember someone from the leaders of the last championship, showed it. And, unsurprisingly, began to see the logic in his reasoning.
Without strong advice try to think about "nonsense" at the weekend - analysis of price series and optimal yield without regard to time.
P.S. About the advisor. It conveniently confirms some seemingly obvious theoretical inferences. And it does not carry out any trading activity.
getch писал(а) >>
Вы анализируете временные ряды, где n - это время. Это концептуальная ошибка. a(n) - должно быть значением цены локального экстремума (ЗигЗаг), либо значение цены через равные накопленные фин. объемы торгового инструмента.
Exactly! Not necessarily a class zigzag - anything which is tied to an established (stationary) process - here to volatility. There is no point in RRR etc in time discretisation for practice.
I wonder if everyone is analysing time-based price series...
As you can see, no. I've been saying that for a long time, but the societal desire to predict the actions of the Fed (for example) based on BP has eclipsed common sense.
You people are stubborn and no wonder. You are incapable of questioning and looking a little differently yourselves at things that have held true for you.
For some reason, logic does not work. One only starts to listen when a bank statement with N-digit sums not in rubles is shown. Doubts and mistakes are not a sign of dementia, but rather the opposite.
Not bad mate training and skills are applied in vain. Many of the results shown here, which are academically unpromising, are successfully applied in trading. And it's not a matter of mathematical training, misunderstanding of terms, etc., but in the initial incorrect applicability and interpretation of the data.
I will not show screenshots of bank statements. I remember someone from the leaders of the last championship, showed it. And, unsurprisingly, began to see the logic in his reasoning.
Without strong advice try to think about "nonsense" at the weekend - analysis of price series and optimal yield without regard to time.
P.S. About the advisor. It conveniently confirms some seemingly obvious theoretical inferences. And it does not perform any trading activities.
A lot of words.
Why are you running around with this stuff like a fool with a swear stick. These questions have long been chewed up and chewed over. And no one's clinging to time here, including Neutron. You had better, dear, try a little harder and figure out what he is writing. Otherwise you get the impression that you do not understand and do not want to understand what it is all about, and think only about what a brilliant idea came to you. And there's only one argument - your 3KB childish primo.
Without strong advice, try thinking about "nonsense" at the weekend - analysing price series and optimal returns without regard to time.
Multicurrency analysis requires at least data synchronisation. Which leads to the requirement to consider time.
Time in general as a part (aspect) of reality exists insofar as there are interactions-interactions.
So-.... >> eternal and timeless good luck to you. ;)
Multicurrency analysis requires at least data synchronisation. Which leads to the requirement to consider time.
Time in general as a part (aspect) of reality exists insofar as there are interactions-interactions.
So-.... good luck forever and ever. ;)
Multicurrency analysis only requires data synchronisation if you are analysing timeframes. If you are analysing a stream of price data from multiple data sources, then no synchronisation is needed for multi-currency analysis. I can explain it better in a private conversation...
A lot of words.
Why are you running around like a fool with this trivia? These questions have long been chewed up and chewed over. And no one's clinging to time here, including Neutron. You had better, dear, try a little harder and figure out what he is writing. Otherwise you get the impression that you do not understand and do not want to understand what it is all about, and think only about what a brilliant idea came to you. And there's only one argument - your 3K childish primo.
Honestly, a lot of nonsense from educated people. They gave you an apparatus, they didn't teach you how to use it. No one here is engaging in self-aggrandizement.
When timeframes are considered, it's a "cling" to time.
I've tried it without persons and for several pages I was proving to forum users (without names), who know MathCad, their blunders in mathematical equation. I was faced with stubbornness which I managed to overcome only with numerous posts, substantiated with MathCad results. I don't want to waste any more time on this.
Certain results seem more meaningful to you when you cite screenshots from mate. packages and use mate. terminology.
They tell you the case and you call it nonsense. You could at least question it a little bit.
I am fiddling because I want to see sensible ideas in market analysis. Which can be, if you go the right way.
Multicurrency analysis only requires data synchronisation if you are analysing timeframes. If you're analyzing a stream of price data from several data sources, the multi-currency analysis requires no synchronization.
You can agree with that. But I don't think it's necessary at all. ;)
Just like you don't have to disagree. :)
Getch wrote >>
Would have explained it more clearly in private...
So what's the problem? I've been on standby for the past hour on both my ICQ and Skype...
:)
I do it because I want to see reasonable ideas in market analysis. Which may be, if we go in the right direction.
"Right" ways are many: the price is an almost perfect random process ("savvy" people, please don't cling to the non-strictness), in which almost any regularity can be found on a finite time interval. Or are you going to say that there is no worthwhile system that uses time?
It's not about time per se. Likewise, you can latch onto any other parameter and say that its use is unreasonable - timeframe, pair, smoothing period, etc.
Any arbitrary parameter in the system is not good. But that doesn't mean that you can't get rid of the system's dependence on this parameter by applying functions from it. You can if you don't cling to its fixed value.
In systems that use price timeframes, you can achieve this by changing the TF in a wide range.
And if, say, the first version of your system uses a muving with fixed period, try to modify the system so that this fixation does not happen by changing this period. In other words, by using functions from different periods, you can ensure that the period dependence effectively disappears in the final version of the system.