Correct calculation of currency indices. - page 13

New comment
 
I have not foundCorrect Currency Index Calculationin the "Correct Currency Index Calculation"thread , does anyone have any suggestions?
 

Catch!

MQL indexing.

 
Strange, I've seen other formulas, with roots.
 

Alexei, this is a question of whether the arithmetic mean or the geometric mean is better.

For indices, there is not much difference and both give an average error of the same order of magnitude. And in general, it depends on the type of NE distribution. Sometimes you have to choose.

P.S. It's a nightmare what's going on in the next thread...

 

I would not have responded there, but I have read it with interest, since its inception and after its reincarnation, too.

The market is typical of the market... The market is not the only one, but it is the only one.

 

If it's a matter of business, then all the flooding should be scrubbed and only the technical stuff should be left, even with emotion (in a reasonable amount). It really is a great thread, isn't it?

2 Neutron: Yeah, well, the difference is about the same as if you calculate returns as differences - or as logarithms of ratios. The second seems ideologically more correct, but the first will do.

 
Mathemat:

2 Neutron: Well, yes, the difference is about the same as if you calculate returns as differences - or as logarithms of ratios. The second seems to be ideologically more correct, but the first will do.

For the first case I was able to mathematically rigorously obtain this expression. That is why I use it. I was even able to estimate the error between the found value and the exact (but unknown to us). This error at seven summands in the Eura index does not exceed 20 points. The error can be reduced by half by constructing in the same way the index for Bax and defining all the necessary indices through it. Then, it is necessary to use the average of each index found separately through EURx & USDx.
 

I decided to practice a little on the hourlies

Took the dollar index quotes from the terminal (DX). Gift #1 - missing quotes. Shown in the table is every 3 hours missing

Calculated DX_Culclate using the formula

=50,14348112 *(1/B2)^0,601*F2^0,142*(1/C2)^0,124*D2^0,095*E2^0,038

As there are only five currency pairs in the terminal and 6 are required, the Swedish krona is scattered evenly among other pairs whose coefficients were taken from the dollar index definition.

The last column shows the unlinking

Conclusion: DX is a traded index and its value does not coincide with the calculated ones.

The question remains: which characterises the currency pair better - the traded index or the computed one?


 

What do you mean by characterisation?

I've decided that it's better to use the formula, the conclusion of which is clear and unquestionable for me. This has been done. This representation of the index - 50.14348112 *(1/B2)^0.601 ... is killing me - I don't understand what's here and where it's coming from.

As for the use of indices in trading, that is a whole topic and as I have learnt, indices do not give a stat advantage over DC commission.

 
Neutron:.

it's killing me - I don't understand what's here and where it's taken from.

This is the formula used to calculate the index

1...67891011121314151617181920...25
New comment