Zero sample correlation does not necessarily mean there is no linear relationship - page 20
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
No, your reasoning is wrong. We know that there is no such thing as below zero, but if we postulate normal for prices, they could theoretically become so if we start modelling the development of the quoting process. So to remove this theoretical possibility, mathematicians get twisted with lognormality. By the way, the tails of the lognormal near zero and at infinity are fundamentally different.
People already know that in fact lognormal does not describe the process. LTCM collapse confirmed it :)
It is not possible to compare two Assets with prices of 1 cent and $400 apiece, but it is possible to compare their logarithms, as they will be separated by a mere constant, removing which will give, for example, their historical graph on the same scale.
No, your reasoning is wrong. We know that there is no such thing as below zero, but if we postulate normal for prices, then they could theoretically become so, if we start modelling the development of the quoting process. So to remove this theoretical possibility, mathematicians get twisted with lognormality. By the way, the tails of the lognormal near zero and at infinity are fundamentally different.
People already know that in fact lognormal does not describe the process. LTCM collapse confirmed it :)
You're complicating things. As has been pointed out, distributions have nothing to do with it at all, none at all. Simply for small price increments: log(P(t+1)) - log(P(t)) ~ P(t+1)/P(t) - 1, where P(t) is price. So logarithms are returns.
No, your reasoning is wrong. We know there is no such thing as below zero, but if we postulate normal for prices, they could theoretically become so if we start modelling the development of the quoting process. So to remove this theoretical possibility, mathematicians get twisted with lognormality. By the way, the tails of the lognormal near zero and at infinity are fundamentally different.
People already know that in fact lognormal does not describe the process. LTCM collapse confirmed it :)
You're complicating things. As has been pointed out, distributions have nothing to do with it at all, none at all. Simply for small price increments: log(P(t+1)) - log(P(t)) ~ P(t+1)/P(t) - 1, where P(t) is price. That is, logarithms are returns.
Why do you need a logarithm that looks like increments when you have the increments themselves? If you want to compare different assets, use percentages.
Why do you need a logarithm that looks like increments when there are increments themselves? You want to compare different assesets, take the percentages.
The need for logarithmetic is partly explained in this post.
Before comparing assets (any number) it is necessary to bring them to the same scale. The way to find maxima and minima for each asset on each window, then transform is theoretical bullshit that has nothing to do with practice. And here's why:
In mathematics, when building theories and solving practical problems, one moves from multiple multiplications (divisions) to additions (subtractions) of logarithms.
As a practitioner, I can say that the recently written correlation indicator (only two financial tools) would not have been possible if logarithms had not been used. Simply, without logarithm, the optimization of the algorithm couldn't be done. And it really is the only correlation indicator that calculates QC almost instantly for hundreds of thousands of sliding windows of any length.
With a large sliding window without using logarithm the supercomputer would always be inferior to a simple solution in MQL4. And this is only for the elementary case of two symbols. And when hundreds of symbols must be compared, each time calculating the covariance matrix. Without logarithm the problem simply will not be solved due to a lack of computational resources. And if you compare Assets in a non-standard way, using numerical methods (for example, solving the quadratic programming problem), the solution will require even more computational resources.
You want to compare the results of your theoretical bullshit with logarithmic approaches, do it. There will be no difference. Only you won't be able to compare hundreds of thousands of results, because you won't be able to calculate them physically.
Moreover, on this forum nobody took relative increments in QC calculations (with QC started the discussion), they took absolute ones. Which, of course, is fundamentally wrong. To take relative ones is suicide, for the reasons mentioned above. That's why it was suggested to do preliminary logarithm.
P.S. I know for a fact that you will stick to your opinion. And this is neither bad nor good.
At one time, having a logarithmic ruler in a man's possession was an indication of his ability.
Now it's just calculators...
;)
The need for logarithmetic is partially stated in this post.
I have already expressed my doubts there that you misunderstand the essence of statistics and statistical relationships - I will repeat the same here - there is no justification there, just amateurish fantasies on near-mathematical topics. You yourself have invented a problem, far from reality, and solved it in the way you know how.
And by the way, you have a gross mistake there in the figures. What you call "graphs with zero MO, one variance and zero correlation" are not. That is, you already have an error after data conversion - you can look no further. The same applies to your recirculation.
As a practitioner, I can tell you that the recently written correlation indicator (only two financial tools) would not have been possible if logarithmetic had not been used.
.I will say more, your correlation indicator is inherently wrong. You have simply substituted the solution to one important problem for the solution from another problem. You've twisted it.
Moreover, on this forum no one took relative increments in QC calculations (with QC the discussion began), absolute ones were taken. Which, of course, is fundamentally wrong.
.Thank you, I laughed. The problem of identifying the correlation between financial indicators lies on a different plane altogether.
P.S. I know for a fact that you will stick to your opinion. And this is neither bad nor good.
.what is there to do? I cannot compromise my principles and share the opinion of a dilettante, pathetically (see http://lurkmore.ru/%D0%9F%D0%B0%D1%84%D0%BE%D1%81) pushing his own delusions (see http://lurkmore.ru/%D0%A4%D0%BE%D1%80%D1%81%D0%B8%D1%82%D1%8C).