Not the Grail, just a regular one - Bablokos!!! - page 57
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Let's, again, how the figures are derived. a sample of 6,736 bars is taken. a window = 236 bars is taken. Move this window from left to right. calculate the unit root test and write in the outermost, 236, place. We get 6736-236 measurements. We see a variable value of the unit root test. The coefficients naturally change.
So what's the question?
Well, I just mean, does it all give a practical result in the form of a profitable system? Well, theory in theory, but in practice it often turns out to be not so bright. In your topic you have demonstrated hypothetical deals only within the window in question. And how does it look in dynamics when the sample is constantly changing? Do you have any tests on the history of, for example, half a year?
I really don't know much about this topic yet, I'm just starting to dig. I have no special education in mathematics, so it's certainly not easy to get into it all. So I'm trying to understand, is it worth spending time and effort, is there a real return on all this? I mean the construction of a more or less stable system in the long term.
Just look at the examples available here, if someone manages to earn on stat.arbitrage, it looks very short-term (for 2-3 months), ie such a feeling that just got lucky, caught a wave. And then there is a long-term stagnation or drain. Although, it would seem, by definition, statistical arbitrage should be quite stable. Or am I wrong?
If there is any cointegration between the two pairs, it is only for a short time interval. Moreover, in most cases the resulting synthetic is almost the same as the cross between these majors (the difference is insignificant). Is there such a thing as a cross hanging in a stable flat for 3 years?
Who says that the pairs have to be combined in such a way that they are quasi-stationary? What if it's the other way around - as unsteady as possible?
Aleksandr recommends that the correlation of traded pairs should be between 35-75%.
Before recommending it, you need to understand what correlation is.
Aleksandr recommends a correlation of traded pairs between 35-75%.
I've seen his other recommendation of 60 (or 65 I don't remember exactly) - 85%
0.60-0.85 already makes sense, as opposed to 0.35-0.70
Well, this is debatable. In fact, the correlation can be normal (I do not know how to put it more clearly), i.e. for example on some TF and with some certain sampling window depth the correlation between FI usually lies within the range of 0.75-0.85. But there are moments when the correlation breaks down and becomes abnormal, say 0.3 or even changes its sign.
Therefore, recommendations can be divided into two types:
1) selection of FI to work with (say, KK=0.6...0.85)
2) Trading signals. There are variations. After all, if the correlation is abnormal, it will eventually come back to normal. You can use it (probably, at least I tried it - it works if the correlation is not on flying FI) and enter at its low values, hoping for a return to normal. QC
So, I was talking about recommendations for FI selection.
An illustration of one method of using QC. There is no direct Pearson calculation, but the principle is the same. There is no need to wait for slamming. The disadvantage is the flatness that swings the FI lot.
Well, I just mean, does it all give practical results in the form of a profitable system? The theory is theory, but in practice it often turns out to be not so rosy. You have demonstrated hypothetical deals in your thread only within the window in question. And how does it look in dynamics when the sample is constantly changing? Do you have any tests on the history of, for example, half a year?
I really don't know much about this topic yet, I'm just starting to dig. I have no special education in mathematics, so it's certainly not easy to get into it all. So I'm trying to understand, is it worth spending time and effort, is there a real return on all this? I mean the construction of a more or less stable system in the long term.
Just look at examples available here, if someone manages to earn on stat.arbitrage, it looks very short-term (for 2-3 months), ie such a feeling that just got lucky, caught a wave. And then there is a long-term stagnation or drain. Although, it would seem, by definition, statistical arbitrage should be quite stable. Or am I mistaken?
Read what you are replying to. It says 6,736 bars (year) by which time the 236 bar window shifts.
Bloody hell! If you don't read the posts, then don't respond to them.