Forget random quotes - page 64

New comment
 
C-4:
It is necessary to do a test on known dependent data. TA indicators without lag, e.g. RSI are very good. It is known that it is not the price that depends on the indicator, built in our terminal, but the indicator depends on the price. The granger test must show exactly this relationship: with a very high probability, rsi depends on price.

I don't see the point in checking Granger. You can see from the graphs above. You can't see the correlation for the whole sample, but for 30 observations there is something.

I typed an article on the indicators. There are other methods besides Granger.

 
faa1947:

I don't see the point in checking Granger. You can see from the graphs above. You can't see the correlation for the whole sample, but for 30 observations there is something.

I typed an article on the indicators. There are other methods besides Granger.


This is about Granger itself. We need to see how effective it is. We don't know anything about the series we're studying, we put Granger on them, we get some result. Can we trust the method and therefore the result? We can, if we make sure that it correctly determines the dependences, which we know in advance.

p/s/ And in general, in research, the rule of thumb method clearly does not work. We need to have a good understanding of the nature of the data and the mechanics of the methods we use to analyse that data.

 
C-4:

It's about the Granger itself, we have to see how effective it is. We don't know anything about the rows we're studying, we put Granger on them, we get some results. Can we trust the method and therefore the result? We can, if we make sure that it correctly determines the dependences, which we know in advance.

Flag on the playbill and I won't be checking out the Nobels. There are so many interesting things that are simply eye-opening, but that I don't get around to.

Thank you for the material, which has broadened my horizons.

 
C-4:


In fact, the poking method doesn't work in research. You have to understand the nature of the data and the mechanics of the methods we use to analyse the data.

This "researcher" only prefers the poke method! Well what about.... nobel... you don't have to think about anything... poke... poke... poke...

And the fact that this poking is useless, doesn't bother our "researcher".

;)))

 
C-4:
It is necessary to test on data which is known to be dependent. TA indicators without lag, such as RSI, will work very well. We know for sure, that not the price depends on the indicator, built in our terminal, but the indicator depends on the price. The granger test must show exactly this relationship: with a very high probability, the rsi depends on the price.
Or take a series, shift it by 1 step and check how it will depend :) and the dependence there is ironclad!
 
faa1947:

I would like to add that in statistics any abstract truths are questionable. You always have to look at the conditions of application, in statistics more than anywhere else the devil hides in the details.

If about Granger, we need to clarify the word "linearity", which comes of two kinds: linearity in the parameters we are estimating, and linearity in the variables, non-linearity of which is usually not taken into account, as it is removed by substitution.

I'm not sure, but it seems to me that causality cannot be considered without the model itself being used for these variables. If our model is linear in parameters and its parameters are estimated by MOC, then the Granger test can be trusted, if the model is non-linear in parameters, then the Granger test probably (I am not sure) cannot be trusted.

Mindful of your recent discourse towards physicists and mathematicians, I note that the physical model is still primary. Without it, you can spend six months figuring out which variable substitution produces the result. The problem is that information, through which the "Granger causality" (notice, information again!) is defined, is in itself a thing similar to harmonic signal in circuit theory: get it in linear system, and calculations are quite feasible; but as soon as nonlinear system appears on the way of information flow, it starts to happen, meat grinder, from which extracting useful component, not knowing the properties of system, is almost impossible.
 
avtomat:

This "researcher" only prefers the poke method! Well what about.... nobel... you don't have to think about anything... poke... poke... poke...

And the fact that this poking is useless, doesn't bother our "researcher".

;)))

Unfortunately, I get the same impression. Faa1947 did some tests. Got some vague results. The nature of these data remains unknown. The idea of these "Nobel tests" is also not quite clear, even in my opinion faa1947 does not quite understand how they are arranged and what exactly are they looking for. In the end, nothing new was found out. Why these tools were chosen out of hundreds of Toolsbox EViews is still not clear to me. The aims of the research were not defined but it is the aims that determine which tools we should use.

In the end, it was a scientific method with uncertain results.

 
C-4:

Unfortunately I get the same impression. faa1947 did some tests. Got some vague results. The nature of these data remains unknown. The idea of these "Nobel tests" is also not quite clear, even in my opinion faa1947 does not quite understand how they are arranged and what exactly they are looking for. In the end, nothing new was found out. Why these tools were chosen out of hundreds of Toolsbox EViews is still not clear to me. The aims of the research were not defined but it is the aims that determine which tools we should use.

In the end, it was a scientific method with uncertain results.

But with a serious look on your face.
 
C-4:

The result is a scientific poke method with uncertain results.

Reshetov 04.08.2012 17:27

But with a serious look on your face.

Under the influence of these two posts I first had an anecdote. Here it is.

.

.

Hearing about China's successes, two economically active citizens took their last penny and flew to China. They fly in. They listen and one of them says:

- "I don't get it.

And the other one remarks:

- Before the trip, I heard that there are a billion and a half of them.

Both of them, together:

- A billion and a half idiots! And all with serious faces!

 
faa1947:

Under the influence of these two posts, I first had an anecdote. Here it is.

.....

- Wow, a billion and a half idiots! And all with serious faces!

Faa, write more jokes. I mean it. You're pretty good at it. And by the way, it's very rewarding. Actual training in shifting the context of perception.

// And you can finally throw your "serious face" in the trash too. It's not useful in life, but it does a lot of harm. It's useless. Smile more often, it suits you.

1...57585960616263646566
New comment
Reason: