Econometrics: why co-integration is needed - page 24
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
I wouldn't generalise like that. By and large they relied on an efficient market. And when the market had a memory, it all came crashing down.
That is, while the market was in a daze, it (the market) was effective. As soon as it (the market) acquired memory, everything collapsed. It is also interesting, how did it (the market) suddenly acquire memory? There was no memory, and then it suddenly appeared out of the blue --- find out how to generalize this point
:))))))))
Where the hell are you getting your bullshit from?
That is, while the market was in a daze, it (the market) was effective. As soon as it (the market) acquired memory, everything collapsed. It is also interesting, how did it (the market) suddenly acquire memory? There was no memory, and then it suddenly appeared out of the blue --- find out how to generalize this point
:))))))))
Where the hell are you getting your bullshit from?
Well, what can I say, you think that cointegration, unit roots, etc. are cool, it's up to you. My job is to warn you that everything in this subject is not as cool as the books say. You need to understand where the black swans may sit in this methodology.
faa1947:
Where?At least in those assumptions (in English) which limit the applicability of your tools. And also the fact that on forex those assumptions are not substantially met. At least for Granger and unit root I can almost certainly tell you that.
Sansanych, I am impressed by your ability to skillfully use "advanced" statin tools, so to speak, without going into too much detail: "A hammered screw holds better than a screwdriver-turned-nail!" (c)
I wouldn't generalise like that. By and large, they relied on an efficient market. And when the market had a memory, it all collapsed.
I specifically looked into this issue - LTCM was trading spreads. The losses, which is what ruined the $100 billion fund - came from arbitrage trades. What is there to talk about, the group at Solomon Brothers from which this fund grew was called that, the domestic fixed income arbitrage group.
The reliance on an efficient market, yes, they had that, but the reliance there was from the opposite direction. They exploited inefficiencies in an almost efficient market.
The funny thing is, I looked on the forum of broker A., it turns out there are also fans of statistical arbitrage. For many hundreds pages they discuss arbitrage averaging methods there. They don't know that it was averaging that was one of the steps to the scaffold for LTCM. This is to the question of black swans. Plus, the black swans are in the wrong assumptions about the applicability of the methods.
I looked into this issue specifically - LTCM was trading spreads. The losses that ruined the 100 billion dollar fund came from arbitrage trades. What is there to talk about, the group at Solomon Brothers from which the fund grew was called that, the domestic fixed income arbitrage group.
A reliance on an efficient market, yes, they had that, but the reliance there was on the opposite. They exploited inefficiencies in an almost efficient market.
The funny thing is, I looked at the forum of broker A., it turns out there are also fans of statistical arbitrage. For many hundreds pages they discuss arbitrage averaging methods there. They don't know that it was averaging that was one of the steps to the scaffold for LTCM. This is to the question of black swans.
Experience shows that all brilliant ideas perish on small things. There were options in this fund. Scholes couldn't do without them. Can we compare? Indeed, should we?
There is a concrete idea to use cointegration. There are a number of tools and evidence in this area. Grail? I don't think so. But it's much more interesting than two mashups.
So far I've found a bug in my optimizer (it's a homebrew one) that was showing incorrect results. I'll correct it, and then we'll see.
I keep trying to bring the collective together to discuss specific problems. It's not so simple in cointegration. Just look at the graph of the probability value on the unit root test.
But, it doesn't work.
At least in those assumptions (assumptions in English) that limit the scope of your tools. And also in the fact that on fore, those assumptions are not substantially met. At least for Granger and unit root I can almost certainly tell you that.
And more specifically. What are the assumptions that are limiting?
What assumptions are not met?
That's interesting to me.
Experience shows that all brilliant ideas perish on small things. There were options in this fund. Scholes couldn't do without them. Can we compare? Do we have to?
Is it clear that the fund was trading the spread between options and other instruments? Is it clear that the options calculation problems have been multiplied by the arbitrage trading problems?
Yes, of course.
If you take cointegration, even changing the method of estimating cointegration regression leads to different results. That's what I mean. That's the kind of thing I want to get into.
And more specifically. What assumptions that are limiting?
What assumptions are not met?
This is interesting to me.
The simplest thing that follows from the principle of construction of both tests is that the residuals of regression equations included in the tests must be stationary and uncorrelated with the series itself, otherwise the method loses its meaning. For Granger - all of the above, but for any number of lags in the equations (which in practice is generally difficult to implement - that's why this test is good primarily for macroeconomic data, where the length of series - annual, quarterly, monthly - usually maximum dozens of samples, but not millions)
And a lot of other subtleties.... The normality of the residuals distribution, for example... (also not very fulfilled)
Plus, as far as causality is concerned, Granger introduced an excellent definition of it, but like any ideal, such a formulation has proven to be unverifiable in practice. So the test of the same name, even if all the prerequisites are fulfilled, will surely only show you the absence of causality if it really does not exist, but not its presence if it really does.