Statistical arbitration - page 3

[[Deleted]]

papaklass:
Alpari.

except the classic account that has this thing is an atavism of their kitchen past =(((

[[Deleted]]

I haven't seen worse quotes than Alpari from any broker (currencies, not interested in other things).

[Alexey]

I have no idea how to form an automated portfolio, because manual creation of hundreds of instruments is not so cool. And all parameters of the position can be coded in Magik, thanks to the big number in five.

[Anatoli Kazharski]

ivandurak:
Has anybody got any idea how to form a portfolio automatically, it's not nice to fool around with hundreds of symbols manually . I think all parameters of the position can be coded in Magik (it's great in five).

On the fourth forum, Reshetov seemed to have made a very interesting presentation on this topic.

ivandurak:
... Besides, all parameters of the position can be coded in magik (the 5 has a large code).

At first glance it seems like an interesting approach. To store several values at once in one value. ))

[Alexey]

tol64:

On the fourth forum, Reshetov seemed to have made a very interesting presentation on the subject.

If you meanhttps://www.mql5.com/ru/forum/123919 Not quite the same, there are fixed instruments, I suggest to form two portfolios automatically, one for buy and another for sell. So the portfolio may contain from one to ..... instruments, besides the weight of the instrument in the portfolio is also unknown.

[[Deleted]]

ivandurak:
I have no idea how to form a portfolio automatically, because it's not nice to manually process a hundred instruments.

IMHO it's a bad idea to do it automatically. Some tools go together and if you moose, you'll take several at once. That's why forming a portfolio is a manual job, no robots.

[Alexey]

220Volt:
IMHO it is a bad idea to do this automatically. Some instruments go together and if you moose, you will take on several at once. So building a portfolio is a manual job, no robots.

Sort of like devircing (rightly so) is designed to reduce risk. In this case we have 2 portfolios where instruments not only don't correlate but also have differently directed movements. Imho you are wrong. There are scientific papers on this topic with Nobel Prizes, if I am not mistaken again.

[[Deleted]]

ivandurak:
It seems that diversification (which is the right thing to do) is intended to reduce risk. In this case we have 2 portfolios where the instruments are not only not correlated, but also have differently directed movements. Imho you are wrong. There are scientific papers on this subject with Nobel Prize winners if I am not mistaken.

If I understand correctly, you're still talking about pre-selection of uniquely going instruments - after all it's hands on. And about the different directionality I think it is not an indicator, the direction can be different and the pulses go at the same time, but in different directions. But of course all this is my IMHO. And in general, it is correct, as our system considers.

[Alexey]

220Volt:
If I understand correctly, you are still talking about a pre-selection of uniquely going instruments - after all, it's manual.

There are two indices: MICEX and RTS, it is clear that they correlate close to 1. In one moment the futures values for the indices go up to say 5%, although the average value is 0.5%. We buy one futures and sell another one, and when they meet again we make a reverse operation and take a profit. It is a simple example that lies on the surface and you can not fix it in time and use it to your advantage, because there are market makers. On the other hand no one prevents you to form your own portfolio (read index). Now count the number of possible portfolios consisting of 5 instruments from 36 possible, at a glance about 300000. I have not dug in this direction and perhaps there is a rational basis.

[[Deleted]]

ivandurak:
There are two indices - MICEX and RTS, it is clear that they correlate close to 1. In one moment the futures values of the indices go up to say 5%, although the average value is 0.5%. We buy one futures and sell another one, and when they meet again we make a reverse operation and fix profit. It is a simple example that lies on the surface and you can not fix it in time and use it to your advantage, because there are market makers. On the other hand no one is stopping you from forming your own portfolio (read index). Now count the number of possible portfolios consisting of 5 instruments from 36 possible, roughly about 300000. Try so many combinations manually, I can hardly imagine how. I have not dug in this direction and perhaps there is a rational reason.

My position is the following: for example we create two portfolios 1 and 2, the composition of the portfolio number 1-5 pairs of instruments (in one pair iinstrument for a sell and buy), each pair follows the behavior of another pair (if the first pair of divergence increases, then the other pair of divergence increases), i.e., the correlation between the pairs is high. Portfolio composition #2-5 pairs of instruments, each pair does not repeat the behavior of the other pair. Then we enter the market with our portfolios on different accounts (the average divergence on #1 - 5%, on #2 - 5%). We look at the next day - in the portfolio #1 - 15% average divergence (everything went down), and we lost money, in the portfolio #2 - average divergence is approximately the same (the movements are mutually compensated). Conclusion - one well-thought-out portfolio rather than a pile of repeating instruments, and if there is not enough - increase volume of transactions, rather than the number of instruments. Drawdown is present in any system, and its depth depends on diversification competence.