Correct calculation of currency indices. - page 8

 
neoclassic >>:
Urain, интересуюсь, на практике индексы действительно проще прогнозируются чем пары?

The answer is ambiguous. On the one hand it is true and the forecast is more often justified by the indexes.

On the other hand, because of the poor quality of the forecast, it is not possible to determine the direction of the pair properly.

I.e. index forecasting is not an end in itself, but only a way to determine the direction of a pair.

It turns out that we look at the indicator and say that the indices will move in the opposite direction, at that the angle of slope of the pair's base is smaller, it means that the pair will go up and the indices move in the similar way, but the pair doesn't go up and goes sideways.

And the forecasting error is to blame.

I think this way is good only for visual decision making, though I have planned it for exact calculations.

 
Hm, thank you. What if we try to trade only indices, e.g. get a forecast to buy euros and buy euros against all currencies?
 
neoclassic >> :
Hm, thanks. What if I try to trade only indices, for example if I get a prediction to buy euro and I buy euro against all currencies?

Trading in a basket is hedging

hedging is aimed primarily at capital preservation rather than capital appreciation

poor spreads for some instruments are not conducive to multi-currency trading

I don't have a lot of lots that would allow me to make exact percentages of the basket unless the deposit is really big.

 
neoclassic >> :
Hm, thanks. What if we try to trade only indices, e.g. we get a forecast to buy euro and we buy euro against all currencies?

No, we don't buy the euro against all currencies, but only against the most overbought, i.e. the weakest against the euro. Otherwise, as sab1uk said, it would just be a hedge.

 
neoclassic >> :
Hm, thanks. What if we try to trade only indices, for example we get a forecast to buy euro and we buy euro to all currencies?

There's a thread to this, I think we need to try and pull it.

Here I am re-posting the refined indicator (as promised).

Files:
 
Alex5757000 >> :

No, we do not buy the euro against all currencies, but only against the most overbought ones, i.e. against the weakest euro. Otherwise, as sab1uk says, we obtain a hedge.

overbought, oversold, it's all relative. An analogue of a hedge would be if we buy eurusd and sell eurgbp. In our case we buy EUR to all currencies included in the calculation of the index.

sab1uk >> :

trading in a basket is a hedge.

hedging is aimed first of all at capital preservation and not at capital appreciation

bad spreads for some instruments are not favourable for multicurrency trading

Lots do not allow you to calculate the exact proportions of the basket in percentages unless the deposit is very large.

It is not hedging, it is buying an index. And the calculation of the exact proportions is a real problem.


 

I tried simulating the calculation of the currency index by initially setting currency movements by random numbers and then calculating conditional cross-rates. Then I calculated the index.

E - initial curve, Er - by the equation EUR*USD*CHF*JPY*GBP*CAD*AUD=1. Em, Eb, Ex - according to the equation EUR*USD*CHF*JPY*GBP*CAD*AUD=K. The K coefficient has been calculated in 3 different ways.

I set the relative movement of notional currencies as a random number +-2.5% for 3 random currencies, and +-0.625% for the others

 

I once poked around the problem of studying the predictability of currency indices.

I found that the index of a particular currency can be reconstructed with an accuracy of a few percent, i.e. very reliably. This accuracy is determined by the number of instruments included in the same index. Nowadays it is easy to find a brokerage company with up to 5 or more instruments that include the same index, i.e. there are no problems with the accuracy. But there is a big problem concerning the predictability of indices. And it is no better than the predictability of the initial instruments!

Thus, all this fuss about indexes, in fact, reminds a shaman's tambourine dance around a sofa.

 

Neutron, how did you estimate the accuracy of the indices?

Personally for my TS, you need to calculate the sign of the index movement, and here is the problem.

For example, we have 5 currencies, two of them rose three remained in place. Most cluster indicators will show these two currencies increasing and the other three currencies decreasing. But the same indicator will show if 3 currencies fell and two stayed in place. Therefore I have assumed that the currencies with minimal movement on average give an overall movement of about one.

 
Yes, how... I take and generate synthetics, then divide one by another - get aanalogues of cotier and solve the inverse problem of restoring the original synthetics from their relative values (cotier). It is clear that in such numerical experiment, I can generate any necessary number of tools and compare the restored values with the initial one at my pleasure. Now, the recovery error exponentially tends to zero as the number of pairs increases. In fact, it is enough to use 5 pairs of instruments to guarantee the accuracy of the index better than a point. Once again I say that index prediction is not easier and not more difficult than the initial instruments!