Discussion of article "Patterns available when trading currency baskets"

New comment
 

New article Patterns available when trading currency baskets has been published:

Following up our previous article on the currency baskets trading principles, here we are going to analyze the patterns traders can detect. We will also consider the advantages and the drawbacks of each pattern and provide some recommendations on their use. The indicators based on Williams' oscillator will be used as analysis tools.

We should analyze the overbought/oversold levels crossing point. At the time of this writing, it is located at the closing of a three-hour candle 17.10.2016 on H1: 

Author: Andrei Novichkov

[Deleted]  

Well, here we are.

Author, your indicators do not show what is shown in the pictures. Why? Are there any normal indicators that would be realistic to use?

 
Must1980:

Well, here we are.

Author, your indicators do not show what is shown in the pictures. Why? Are there any normal indicators that would be realistic to use?

These are test indicators. Indicators for analysing basket currencies are extremely sensitive to history. And here, please note that in order to analyse one pair, you need a high-quality history for more than ten pairs. After all, two baskets, each of which consists of many pairs, will be analysed. I didn't see the sense in building complex history control tools into a test indicator, the purpose of test indicators is different. But, as you put it, there are "normal" indicators. I wrote about them in the last article, and I myself developed all the indicators I gave as an example. RSI, WPR, SToch, Index, they are all available in quite working versions. But I can't put them on the market, because they are configured for management from a "single centre", which consists of several text files. This is forbidden by the rules.
 
In my opinion, replacing the geometric mean with the arithmetic mean will not give a greater gain in performance, but such an optimisation will spoil the result, from a mathematical point of view. WPR indicator is not a bad choice - it is Stoch with zero delay, but it was more correct to build it according to the results that you get in the indicator testBasket.mq4, that is, first get a "clean" chart of the euro, without the influence of other currencies, and according to those values to forgive the classical WPR, it is not difficult to do if you calculate it yourself, without using the iWPR function.
 
Иван:
In my opinion, replacing the geometric mean with the arithmetic mean will not give a greater gain in performance, but such an optimisation will spoil the result, from a mathematical point of view. WPR indicator is not a bad choice - it is Stoch with zero delay, but it was more correct to build it according to the results that you get in the indicator testBasket.mq4, that is, first get a "clean" chart of the euro, without the influence of other currencies, and on those values to forgive the classical WPR, it is not difficult to do if you calculate it yourself, without using the function iWPR.

I haven't tried to measure the difference in performance, so I can't say anything. Geometric mean can't be applied everywhere. If there are initial values that change sign unpredictably, then in the general case we will get an exception, which we don't need ). That's why the arithmetic mean is used where there is such a risk. I mentioned it somewhere, by the way. I even gave a graph comparing geometric mean and arithmetic mean. I should note that in "combat" indicators I have the ability to change the calculation method, where possible, of course. And I would express my doubts about "spoiling the result". Yes, these methods will show different values in absolute terms. But they will show extremes in the same places. Whether they will show the same moments of exit from overbought/oversold zones - this needs to be checked, I quite admit that the geometric mean will give more accurate data, with earlier exit. But once again, it needs to be checked, to have a statistical base.

As for your suggested way to apply WPR calculation to the "index" (testBasket.mq4 counts it exactly), I think it is a very interesting idea and I will try to implement it if possible. I should note that the description of this method would be beyond the scope of this article.

 
MetaQuotes Software Corp.:

New article Patterns available when trading currency baskets has been published:

Author: Andrei Novichkov

Hi Andrei i am interested with this indicators. can you develop readily applicable for me for EUR/USD and 

some other pairs?

 

Hi, kumarsk
Of course. WPR, or RSI, or STOCH for 28 pairs.  Look at the image for an example (WPR).
Files:
EURCHFH1.png  51 kb
 
make sense
 
In the testWRP indicator the EUR currency is always in the first place, and in testWRPjpy the yen is always in the second place. Will the indicator code for CHF change ?
 

The code is slightly different for each basket currency. For example, you want to buy a basket on USD. To do this, you should enter the sale on EURUSD, because USD is the "second" currency. It is written in detail in the first article (or in the second one, I don't remember exactly) on this topic. There is a table for all basket currencies. I would like to remind you that test indicators are not suitable for use in "real life".

New comment