Econometrics: one step ahead forecast - page 105
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
Predicting ZZ reversals. EURUSD_H4 500 last observations
A fairly decent ZZ with a period of 12
I form a file of the following form:
Where the tops of the reversal are marked 1 and -1. There are only 17 such tops at 500 observations.
I form the regression equation
zz_high eurusd(-1 to -100) c @trend
I will predict a downwards reversal (high points) on the value of the previous 100 EURUSD bars.
Assuming I have a binary model (only two values to which the initial series fits: 0 and 1). The evaluation method: BINARY broken:
Getting the result of the fit:
There is only one error in the fit.
But the most surprising thing is that only a discrete result is fitted to a continuous quote
Now making a prediction
1. On a continuous quotient, reversals are predicted!!!!
2. there is no error. there is no previous 100 bars in the beginning, so there is no prediction at all - it is not an error. Because of discreteness the prediction lies on top of the fact and therefore the fact is not visible
An interesting observation about the team.
New production - predicting reversals
Amazing result.
And not a single post!
Predicting ZZ reversals. EURUSD_H4 500 last observations
A fairly decent ZZ with a period of 12
I form a file of the following form:
Where the tops of the reversal are marked 1 and -1. There are only 17 such tops at 500 observations.
I form the regression equation
zz_high eurusd(-1 to -100) c @trend
I will predict a downwards reversal (high points) on the value of the previous 100 EURUSD bars.
Assuming I have a binary model (only two values to which the initial series fits: 0 and 1). The evaluation method: BINARY broken:
Getting the result of the fit:
There is only one error in the fit.
But the most surprising thing is that only a discrete result is fitted to a continuous quote
Now making a prediction
1. On a continuous quotient, reversals are predicted!!!!
2. there is no error. there is no previous 100 bars in the beginning, so there is no prediction at all - it is not an error. Because of discreteness the prediction lies on top of the fact and therefore the fact is not visible
Do you use the standard sounder? Don't you think that the sounder works coarsely on the graph and you need to increase its sensitivity?
Are you using the standard reed switch? Don't you think that the gating is rough on the graph and its sensitivity needs to be increased?
An interesting observation about the team.
New production - predicting reversals
Amazing result.
And not a single post!
Why should people write in vain? Personally, I am an economist-mathematician-cyberneticist, practicallyan"econometrician". What you have written here has some basis, but for a serious conversation it is NOT SCIENTIFIC enough. There are some knowledgeable people here, though the moderators of this forum have banned a couple of them, such as Privalov. But knowledgeable people need a clearer formulation of the problem and the method, and with a hard check of applicability of initial conditions of the method to the given problem. Otherwise it is possible to puddle for decades.
I can say the same about tens of articles on this site. This is the same say here KNOWLEDGE people, well, for example here in the discussion of a recent article "on statistics".
And yet you colleague are taking on the eurik at once. The Euro is NOT a legitimate currency at all. And its price involves a collection of various flows from 20+ DIFFERENT countries. Therefore it is 20 times harder to predict the EUR than any other pair.
I say this only out of respect for your courage and lack of arrogance that other arrogant article and post writers here suffer from.
Hat-trickers, my ass (that's not you).
Why should people write in vain? I personally am an economist-mathematician-cyberneticist, practically an "econometrician". What you have written here has some basis, but for a serious conversation it is NOT enough science. There are some knowledgeable people here, though the moderators of this forum have banned a couple of them, such as Privalov. But knowledgeable people need a clearer formulation of the problem and the method, with a hard check of the applicability of the initial conditions of the method to the problem. Otherwise it is possible to flounder for decades.
I can say the same about dozens of articles on this site. The same thing is said here by KNOWING PEOPLE, for example in the discussion of a recent article "on statistics".
And yet you, my colleague, are taking on the eurik at once. The Euro is NOT a legitimate currency at all. And its price involves a collection of various flows from 20+ DIFFERENT countries. Therefore it is 20 times harder to predict the EUR than any other pair.
I say this only out of respect for your courage and lack of arrogance that other pushy article and post writers here suffer from.
Shapkozakidateli, damn it (it is not about you).
There is NOT enough science to justify it - you don't need to. What, for example, is a sufficient scientific justification for ZZ? They are lines drawn with a ruler on a graph.
The REASON is to create a profitable TS. This task is the same for everyone on this forum.
The euro is NOT a legitimate currency at all. Consequently, predicting the eurik is 20 times harder than predicting any other pair. - Do you have any studies that suggest this conclusion? Did you draw it yourself? How?
Why should people write in vain? - This is correct! We must simply obtain the results of this method on the demo account and discuss on their basis. If we are going to use this method, we'll see if the indicator will give good results.
The euro is an ILLEGAL CURRENCY altogether. Accordingly, predicting the eurik is 20 times more difficult than predicting any other pair. - Do you have any studies that reflect this conclusion? Did you draw it yourself? How?
Yeah right here, wrote here on the forum in the summer of 2009:
https://forum.mql4.com/ru/24358#187696
"The euro is a non-existent currency of a non-existent state. That's why it's so twitchy."
EU countries don't have a constitution. No constitution - no state. No state - no legitimate money from that mythical state.
It's all these "financiers" who have now realised that all this euro stuff is illegal, and are clucking about "saving" the eurik. They should have just read the MQL4 forum before. now it's too late.
Nothing is going to work for them. Law is above currency. And without law, the only valid currency becomes good old gold (for guarantees, collateral and security of funds) and silver (for small day to day current settlements).
"The euro is a non-existent currency of a non-existent state. That's why it's so twitchy."
Thanks for the post.
But knowledgeable people need a clearer formulation of the CHALLENGE and the METHOD,
I don't really understand the problem, but I'm trying to compensate by using a standard package. I have a very wide range of mating tools. Got the systematicity of the tool instead of trying to apply any single tool.
I have tried to solve the problem of applicability of initial conditions of the method to the given problem.
I tried to solve the applicability problem by decomposing the unsteady quotient into its components.
And thus you colleague take up the eurik at once.
It's not about the euro. I don't see the approach to justification of predictability by the model. Compliance with all the rules when building the model does not give a tangible increase in the probability of success of the forecast - the same 20-30% turbulence
Have you investigated the volatility of different currencies and compared it and found an abnormal value just for the euro? Or how did you come to that conclusion? "Twitching" - is that from an economist-mathematician-cyberneticist point of view, practically "econometrics"?