Arbitration Advisor - page 11

 
Vitaly Muzichenko:

One of my first entries is on youtube, you can see there, the strategy has changed slightly.

The method is as simple as five cents, you only need two things: an analytical mind, and a couple of tools to work with.

You don't have a robot, he wanted a robot)
 
Maxim Dmitrievsky:
You don't have a robot, he wanted a robot.)
Well, I wrote that it takes a lot of work to get something. You cannot make a robot using the system, so you can work with your hands and look for a bot in the meantime. It can last even 10 years, and during that time you can earn something than just being in search for 10 years).
 
Vitaly Muzichenko:
Well, I wrote that it takes a lot of work to get something. You can't make a robot using the system, so you can work with your hands and look for a bot in the meantime. You may work for 10 years, and you may earn something rather than just search for 10 years).
well, you have pair trading, without synthetic analysis, am i right?
 
Maxim Dmitrievsky:
This is not a trading robot, he wanted a trading robot.)
I did not want a robot.
Vitaly Muzichenko:

One of my first entries is on youtube, you can see there, the strategy has changed slightly.

The method is as simple as five kopecks, you only need two things: an analytical mind, and a couple of tools to work with.

I have watched the video.
If I am not mistaken, the charts of different instruments are superimposed on 1 chart by connecting them all in 1 point at some distance from the start of the viewing area (the right edge of the chart). 60 bars apparently. The idea is described in the "Not the Grail..." thread.

I did a similar thing... The problem I saw was that if you move the convergence point to 1000 bars and beyond, you see that the charts may diverge and never converge again (or converge in months or years).

I.e. all the same it turns out to be a "guessing game". Probably, the instruments will correlate for some time, but one day the moment will come when you will place a bet on convergence, and the pairs will not converge any more.

So, I can't do without stops after all. In my case there were StopOuts ))
 
elibrarius:
I didn't want a robot.
Watched the video.
If I'm not mistaken, the charts of different instruments are superimposed on 1 chart by joining them all in 1 point at some distance from the start of the viewing area (right edge of the chart). 60 bars apparently. The idea is described in the "Not the Grail..." thread.

I did a similar thing... The problem I see is that if I move the convergence point to 1000 bars and beyond, I see that the charts may diverge and never converge again (or converge after months or years).

In other words, it is still a "guessing game". Maybe the tools will be correlated for a while, but one day the moment will come when you bet on the convergence and the pairs will no longer converge.

So, I can't do without stops after all. In my case there were StopOuts ))

I mean the starter's topic, he wanted a bot... )

about the charts - yeah, it's pure guesswork, that's why i have my doubts too

 
Maxim Dmitrievsky:

I mean the top starter, he wanted a bot... )

about the charts - yes, it's pure guesswork, that's why i have my doubts too

Well our task - while the instruments are correlated, as much profit on them as possible (maybe there are a few days, until the news or market makers shift one of them to another level).

And to minimize losses at this divergence. How do you predict the moment of divergence? Any ideas? Just do not trade before the news? But they are sometimes quite profitable, if the instruments remain correlated.

 
elibrarius:

Well, our task is to make as much profit as possible while the instruments are correlated (there may be some days until the news or market makers shift one of them to another level).

And to minimize losses at this divergence. How do you predict the moment of divergence? Any ideas? Just do not trade before the news? But they are sometimes quite profitable, if the instruments remain correlated.

I don't play the ifs and buts... The answer is in the question itself - there is no way to predict) Either rigorous backtesting and risk analysis

Many people write here saying they are smart, know R and statanalysis, use Matlab, all sorts of complex terminology. But nobody shows you the results - though you should have started with them.

I've already sent here bot's code for pair trading on correlations and it even earns sometimes but fails in a long run. And there is no need to possess a complex mathematical tool for writing something like that, the correlation and logo deviations are enough, and nobody has learned yet to predict the rest (future).

 
elibrarius:

In other words, it is still a "guessing game". Maybe the instruments will correlate for a while, but one day the moment will come when you bet on convergence and the pairs will no longer converge.

The point is that we take pairs that are relevant at the moment, i.e. we don't use the same pairing. I only once had a hanging portfolio, it closed after 2 months, but the loss was up to 130$ with 0.03 lot.

I have to take pairs and directions according to current situation, then there are practically no dips.

I take history 340-370 bars on М30, practice showed that it is optimal. If prices diverge proportionally then it is possible to exclude them, but if one moves faster, or better by an impulse, then it is an ideal entry.

Profits should be taken in the region of 20pp, if you wait for more, then there is a big chance to overstay, you can take 10-15pp - ideal. If there are news movements then we should not close pairs (portfolios) that have one currency pair but in different directions, then in case of a strong movement the loss/profit will be about zero, after everything has settled down - we close big profit, and then wait for the return, or break-even. It is also possible to add a third pair to the existing portfolio, but not a lot.

I first tried fixed portfolios, the results were not impressive, but I started to select them according to the market and I was pleased with the results.

 
Maxim Dmitrievsky:

well, i don't play ifs and buts.......

who quoted hrenfx?

Have you tried it yourself?

 
Renat Akhtyamov:

Who quoted hrenfx?

Have you tried it yourself?

I don't know who, hrenfx has synthetics, his bot is currently being tested to at least get an idea if it's worth the hassle or not.