Zero sample correlation does not necessarily mean there is no linear relationship - page 21
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
And by the way, there's a gross error in your figures there. What you call "graphs with zero MO, one variance and zero correlation" are not. That is, you already have an error after the data conversion - you don't need to look any further.
All you are saying is unsubstantiated. For my own part - in addition to the graphs, I have also attached the raw data. You can easily check it. And, as stated earlier, QC is not affected by multiplying (bringing variance to one) by a constant and adding a constant (bringing MO to zero). Kindly do a rudimentary test on the data provided and report the results here. Not make unsubstantiated claims of gross errors.
.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. Overblown.
Once again, it's all just chatter. Take an indicator reading anywhere, calculate for it with your QC method and compare. The example above is the zero correlation found from the indicator readings. The screenshot specifically underlines in red that the correlation calculated by Mathcad is indeed zero on this sample:
.Thanks, I laughed. The problem of identifying the correlation of financial indicators lies in another plane altogether.
Get on with it, now that you've started.
...
Moreover, no one on this forum has taken relative increments in the QC calculations (with QC the discussion began), they have taken absolute increments. Which, of course, is fundamentally wrong.
It has already been said to you 1000 times. Why do you say for everyone that no one. Have you looked through my computer and checked all the tools developed by the participants of this forum?
The man built an indicator ... based on it made a trading system ... and is making a profit ... You are saying from this point of view that it's not true ? i am not right at all, it's not true in principle, what else is not true?
Stop thinking everyone on this forum is an idiot and you're the best and brightest.
I've told you 1000 times already. Why are you saying that no one is right for everyone? Have you looked through my computer and dug through all the developments of the participants of this forum?
Because I first searched this forum for QC calculations and made sure that relative increments were not used (results posted on the forum) by you or anyone else. If I wasn't looking hard enough, show me.
It has already been argued several times, and not only by me, why using absolute increments when comparing samples of two or more CERs is wrong.
All that is just unsubstantiated on your part. For my part, in addition to the graphs, I have also attached the raw data. You can easily check it.
I'm sorry, are you stupid?! We're talking about this picture here.
There is no MO = 0 and D=1 here.
It's amazing, but this is the third time I've drawn your attention to this gross error. It's as if you can't understand this simple thing at all, let alone debate it.
There is no MO = 0 and D=1 here.
It is amazing, but this is the third time I have brought this gross error to your attention. It seems as if you are not able to understand this simple thing at all, let alone discuss it.
On the basis of what reasoning do you draw such conclusions about MO and variance! Good thing the first post gives the exact date from which the data was taken, so it was possible to recover:
In this case, as I wrote in the first post, the QC is equal to the average product of the BP members of the sample:
Source data attached.
Logarithms are used to establish explicitly that a quantity with a distribution resembling a normal distribution has a lower bound of zero. In deriving the Black-Scholes formula, it is assumed that the price distribution is lognormal, i.e. it is not the price that is normally distributed, but its logarithm.
ARPSS necessarily includes BP detrending. A distinction is made between additive and multiplicative trends. The latter, of course, are logarithmic before detrending BP.
On the basis of what reasoning do you draw such conclusions about MO and dispersion?!
Very simply, MO and variance exist as a statistical concept for random series, you have a "non-random" series. That is, MO and variance do not exist as concepts for them.
1. Roughly speaking, the sign criterion is broken, about 80% of your data is positive (it's a mistake - improper standardization). People in a neighbouring thread are raving about quantiles - this is just the same thing. And from the basic definition of random series.
2. The 'functional' dependencies are clearly visible.
3. and most importantly, it is to the question of knowing exactly what instrument is being analysed - there is nothing random about these instruments. At least in the representation of the data you have.
4. There is no need to hide your misunderstanding behind matrix packages, understand the basics first. And the basis is simple, the statistical analysis (and calculation of MO) can be subjected to "random" series.
5. If you just took the data as is and I showed you how to do it - that would be closer to the truth.
What the hell is statistical analysis? Do you have any idea what you are talking about? We are talking about sampling, QC counts on a sample. Are you aware of concepts such as sampling MO, sampling variance and sampling QC?
As if they were written for you:
A little bit of background.
Another common misconception is to confuse the concepts of "correlation coefficient" (i.e. a characteristic of a stochastic relationship between c.v.) and "sample correlation coefficient" (an estimate - one of many possible - of the true QC). These are actually quite different things, and substituting one for the other is fundamentally wrong.
Very simply, MO and variance exist as a statistical concept for random series, you have a "non-random" series. That is, MO and variance do not exist as concepts for them.
1. Roughly speaking, the sign criterion is broken, about 80% of your data is positive (it's a mistake - improper standardization). People in a neighbouring thread are raving about quantiles - this is just the same thing. And from the basic definition of random series.
2. The 'functional' dependencies are clearly visible.
3. and most importantly, it is to the question of knowing exactly what instrument is being analysed - there is nothing random about these instruments. At least in the representation of the data you have.
4. There is no need to hide your misunderstanding behind matrix packages, understand the basics first. And the basis is simple, statistical analysis (and MO calculation) can be subjected to "random" series.
5. If you just took the data as is and I showed you how to do it - that would be closer to the truth.
Don't waste your energy. Prival tried to explain to him that ACF is a function, not a number - he failed. That's where you have to start.
5. If you were to bluntly take the data as it is and I showed you how to do it - that would be closer to the truth.
Show me, please: