Zero sample correlation does not necessarily mean there is no linear relationship - page 38
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
Just build it and you can see everything with the naked eye.
What are the differences for? No, no, very blunt and primitive - QC for time-shifted rows.
Useless from a practical point of view. We are trading differences. The trader does not care whether a tugrik is worth two roubles now or three, the main thing is how the price will change in an hour, so it is the differences that should be the object of research, including the correlations between the differences that we should be interested in.
Differences between what and what?
the difference between what and what?
The so-called returns, of course:
return[ i ] = Close[ i ] - Close[ i+1 ]
P.S. Got here ahead of you. Here I have a special case with lag 1. The numbering is the same as in MT4.
This is to ensure that you have at least a little understanding of what the correlation coefficient is, before you talk about any constructive reasoning.
5 points. even 10. The author of this branch has written so many letters and programs, he called them both QC and ACF, but has not figured out what they are and what they are about)). Even indirectly makes fools and idiots all, like he alone knows how to properly calculate AFC, and no one before him did not do and can not do it. In general we are all idiots here )))).
The author of the branch open any university textbook and read.... the title of your branch...every textbook talks about it. Congratulations on your discovery ))
I liked it:
если взять место, куда попадает много очень разнородной информации, в частности, связанной с исследуемым явлением, то если информацию правильно отфильтровать и обработать — то вполне себе будет связь.
потому что у любого явления есть протяженные во времени причины (и следствия).
но не обязательно первое — следствие второго.
например, есть связь между количеством утопленников и количеством съеденного мороженого есть положительная корреляция.
просто потому что летом и мороженое едят больше, и утопленников больше.
но если нам очень уж нужно предсказать количество утопленников, а термомента нет и окно заложено кирпичами, то можно вполне себе предсказывать по потреблению мороженого.
всё со всем связано. только обычно очень сложно уловить связь.
Am I correct in assuming that the results of your study summarise the conclusion that can be drawn:
For profitable trading it is necessary and sufficient to establish the dependence between the traded instrument's increments and something else. This "something" can be the dynamics of ice cream sales, the rates of other instruments or the instrument's increments shifted in the past by some counts.
1. If we deal with analysis of previous movements of the traded instrument and predicting the expected direction of price increase based on it, then it is a classical (single currency) TA. Here it is important to determine the optimal time lag. The larger is this value, the larger is the instrument's movement in points, but unfortunately the less is the correlation coefficient. And vice versa, at zero lag we have correlation ==1, while the move==0 and the profit is equal to their product.
2. Considering a non-traded instrument (for example, a currency index) as the second series, we should build a cross-correlation function and select a time lag that gives the maximum product of the correlation coefficient and volatility of the traded instrument on TF=lag. Thus, we use the third series as a leading indicator. By the way, it is the only way to build a real leading indicator. The MA cannot be considered as such by definition, because it is constructed on the basis of the traded series that has negative correlation between samples in the series of the first difference.
3. Considering another traded symbol as the second series (for example some currency pair), we can do the same as in paragraph 2 and use the second instrument as a leading indicator, or trade both instruments. In this case we consider simultaneous opening of two symbols according to the sign of their pair correlation coefficient (for the zero term of cross-correlation function) built for the first difference of both series (Open[i]-Open[i-1]). This is spread trading.
4. The number of instruments traded as described in Sec. 3 can be increased to any integer number. It is important that the instruments in the basket have a strong pairwise correlation at the zero count of the series of the first difference.
I do not understand the author, what is the advantage of trading several instruments compared to the two-currency basket? The author considered this question in details in the parallel thread. It would be interesting to hear his opinion. The idea is that the use of many correlated instruments in the basket should decrease the risks, but we know that the decrease of the risks decreases the returns, so the question is to determine the functional dependence of the risk level and the expected return on the basket. For strongly correlated instruments it may happen that both of these parameters will be almost linearly dependent (for uncorrelated instruments, risks decrease at the root of the number of instruments faster than returns) and the entire advantage of the basket as a low-risk aggregate instrument may be ethereal.
5. A "second" instrument can also be a second DC. For this purpose, we look for the DC having a nonzero term in the cross-correlation function for the instrument with the same name (more simply, the quotes lags/exceeds ours). We use the quote of such brokerage company as a leading indicator. This is inter-dealing arbitrage.
There are no more ways to make money in the Market.