Absolute courses - page 23

[[Deleted]]

Joperniiteatr:
that's it, we're losing the man.... gold cubes on the avatar went..... why do you let yourself be read so quickly(. Psychological profile detected by 75%

Very on-topic avatar. Rubik's cube of D, E, Y.

[[Deleted]]

Dr.F.:
Forget page 12. There, consider that my one curve was arbitrary :-) But of course that's not true. Now we are talking about the problem I set on page 20.

You set the problem, then it turns out to be not telling, then again adding something new, then the point of guessing, if the rules of the game are fluid retrospectively.

[[Deleted]]

Joperniiteatr:


You set the problem, then you understate it, then again you add something new, then the sense of guessing, if the rules of the game are fluid by hindsight.

Why backwards. I have given you a new original problem. Here: my question: is it possible from EURUSD (I denote by ED) and EURJPY (I denote by EY) to draw curves E, D, Y relative to time-varying benchmarks (i.e. values of D, E, Y in a particular bar in the past) so that they satisfy the known equations ED, EY, DY, while correlating with each other by coefficients close to unity?

I am ready to lay out a solution where it is already 0.99 for any pair of E, D, Y.

[Дмитрий]

You have E0 D0 Y0 different on three graphs. And on the fourth, all three are already =1. So what formula did you use to normalise them to make them so similar? Ahhhhhhhh, I get it, there on each bar the coefficient and offset were manually selected, then the question is why you could not get the correlation coefficient to 1, your way is easier than a penny.

[[Deleted]]

grell:
You have E0 D0 Y0 different on three graphs. And on the fourth, all three are already =0. So what formula did you use to normalise them to make them so similar? Ahhhhhh, I get it, there at each bar the coefficient and shift were chosen manually, then the question is why you could not get a correlation coefficient of 1, your way it's easier than a darned tree.

Yikes, my colleague. E0, D0, Y0 - at least open the file EDY.txt - they are all three in the first line. Next, build the columns. These are actually E, D, Y. But if you divide these columns by E0, D0, Y0 to show them on one chart and you can see them, you will get chart 4. What is the coefficient, what is the bias? What do you mean. I have only one condition: at some bar in the past (144 bars in the past in this case) D=1. So E=ED, Y=1/DY (at that bar). That's it. This is now the benchmark. Parrot. An immutable quantity. We plot D from time in relation to that D0=1. E too. And Y too.

[[Deleted]]

Here you go, colleagues, once again, the same trick.

Watch your hands. I am posting the EURUSD and EURJPY files. Here they are:

[[Deleted]]

I take from them 144 bars from the end. I plot the EURUSD, EURJPY, USDJPY charts and from them I find curves E, D, Y. These are the ones:

As you can see the correlation coefficients between the E, D, Y curves I found are already close to 0.99 on average.

[[Deleted]]

That said:

For those wishing to construct for themselves and make sure these pictures are correct, I am attaching the data file with columns E, D, Y respectively:

[Дмитрий]

Normalised how?

[[Deleted]]

And yes, by the way. Don't let the 0.9999+ correlations confuse anyone. They can be made to equal exactly one. It's just that the calculations will take a very long time. The struggle for those +0.00001 to coefficients is longer in time than calculating what is already shown.