Once again, arbitrage, pair trading. - page 8

 
Maxim Dmitrievsky:

Found another bug in the calculations, now the picture is more realistic:

Foot is running +-100 pips and on the brexit it was -400 :)

Pound should have sold on this screenshot.

Nothing to be afraid of now.

There will be a profitable trade.

In general terms, what is the algorithm to build such a spread, if it's not already a secret?
 
Renat Akhtyamov:

The pound should have sold on that screen.

Now there's nothing to be afraid of.

It will be a profitable trade.

In general, what is the algorithm of such spreads, if it's not a secret?

So I wrote before - multiple linear regression. For each of the instruments we build the coefficients of dependence on all the others.

Well, how do you say, profitable... spreads have widened considerably now

The pound has bounced 180 pips again.


 
Maxim Dmitrievsky:

I've written before - multiple linear regression. For each of the instruments we build coefficients of dependence on all the others

Well, let's just say it's profitable... spreads have widened considerably now.

God bless you too)

 

Maxim has said more than once, and no one realises except him, the spread is recalculated with each bar. That is, the ratios change every bar. And this fact will always cause disadvantages for the entire portfolio (in spreads per trade for each currency pair). Plus it will not give any advantage in the analysis, absolutely none.)

 
Maxim Dmitrievsky:

I've written before - multiple linear regression. For each of the instruments we build coefficients of dependence on all the others

Well, how can I say it's profitable... spreads have widened considerably now.

It should be so that the quotes did not use the case of multiple reversals of deals from Buy to Sell and vice versa.

All in all, everything is great, you can see it with the naked eye.

 
Anatolii Zainchkovskii:

Maxim has said more than once, and no one realises except him, the spread is recalculated with each bar. That is, the ratios change every bar. And this fact will always cause disadvantages for the entire portfolio (in spreads per trade for each currency pair). Plus to all this it will not give any advantage in analysis, absolutely none.)

Of course, but this fact seems to be taken into account, or not?

 
Anatolii Zainchkovskii:

Maxim has said more than once, and no one realises except him, the spread is recalculated with each bar. That is, the ratios change every bar. And this fact will always cause disadvantages for the entire portfolio (in spreads per trade for each currency pair). Plus it will not give any advantage in the analysis, absolutely none.)

It can be solved. It is necessary to find the average ratio for a year or two. If this coefficient is not equal to the quotation of the cross, then everything is normal.
 

here's one for pairing ))))

 
Vitaly Muzichenko:

Good day to you, too)


It's simple, here's a breakdown of everything: and alglib has a mn reg.


 
Anatolii Zainchkovskii:

here's one for pairing ))))


I made a similar one... if only we could find a way to determine the significance for each attribute and form an optimal portfolio on the fly, discarding bad pairs