Econometrics: one step ahead forecast - page 107

 
faa1947:

An interesting observation about the team.

New production - predicting reversals

Amazing result.

And not one post!


because you did some bullshit that you're ashamed to show...

about zz - by predicting zz you're predicting what you're trying to avoid - unsteadiness... you don't have to measure anything... just look at the trough peaks - you can see everything... )))

Demi is trying to get your point across about what you call seasonality (cyclicality)... although he talks about valatility but in your model would work more like cycles... that's why you were talking about zz earlier... but you started predicting it...

 
Vizard:


because I've done some bullshit that I'm ashamed to show...

I'm not ashamed. I'm not ashamed specifically.

 
Vizard:

Demi is trying to get your point across about what you call seasonality (cyclicality)... although he talks about volatility but in your model it would work more like cycles... that's why you were talking about zz earlier... but you started predicting it ...

Tics is nothing at all.

Cyclicality to me is a wave with a variable period, but it's a level wave

If you take the distances between the peaks of the ZZ? Most likely not stationary and very few observations (peaks of ZZ)

 
Vizard:


you were talking about earlier... but you started forecasting it...
ZZ prediction is from another thread: what are we predicting? level, level gain, level slope, level reversals - what are we predicting?
 
faa1947:

because you've made some bullshit that you're ashamed to show...

I am not ashamed. I am not ashamed specifically.

It's not bullshit, it's not bullshit, don't get upset about the criticism.

At least it clearly captures a SCIENTIFIC approach. It makes for a good term paper. Also, on this forum, most physicists, "mathematicians" and especially radio technicians would do well to learn that there are different distributions of what they call "probabilities" out there.

The problem is that you jump through a lot of your assertions-links, considering them completely obvious, or rather considering their applicability to trading in this particular place of your methods to be completely obvious. However, it is not so. In other words: You must PROVE at least EXACTLY why we do this and that, and do that.

No one has a mathematical model of trading (well, let's just say it's not published). No one requires you to STRICTLY justify why we do exactly this sequence of gestures (calculations), but for constructive work, such a non-strict justification is extremely desirable.

More logic, expressed in simple and understandable Russian words - in places of transition from one method to another.

Feuerstein? Not too abstruse and generic?

 
AlexEro:

It's not bullshit, don't be upset by the criticism.

At least it clearly has a SCIENTIFIC approach. It makes for a good term paper. Also, on this forum, most physicists, "mathematicians" and especially radio technicians would do well to learn that there are different distributions of what they call "probabilities" out there.

The problem is that you skip over a lot of your assertions-links, considering them to be perfectly obvious, or rather considering it perfectly obvious that they are applicable to trading at this particular point in your methods. However, it is not so. In other words: You must PROVE at least EXACTLY why we do this and that, and do that.

No one has a mathematical model of trading (well, let's just say it's not published). No one requires you to STRICTLY justify why we do exactly this sequence of gestures (calculations), but for constructive work, such a non-strict justification is extremely desirable.

More logic, expressed in simple and understandable Russian words - in places of transition from one method to another.

Feuerstein? Not too abstruse and generic?

Let's be more specific.

1. Predicting the level - we just want to and the experience of predicting the increment is not good.

2. Verbal model: quotier = trend + noise + periodicity + seasonality. There is no seasonality in forex, I don't know what to do with periodicity. We model the trend + noise.

3. This model allows to distinguish in a number of steps what we can work with, and what we cannot - reduce in absolute value and stabilize.

4. Then we write down a series of regressions.

If you agree with this step, I will continue

 
AlexEro: I only explain it out of respect for your courage and lack of arrogance that other pushy writers of articles and postings here suffer from.

Shapkozakidateli, man (that's not about you).

Hi Alex, very good to see you (as well as all the other "old timers" who came back here anyway). By the way, Prival is here too, but very rarely and under a different nickname.

No hard feelings: we have the same liver as you do.

As for the hat hat-trick, please clarify. If it is me who is so pushy - I will be glad to hear criticism on the case.

Also, on this forum, most physicists, "mathematicians" and especially radio technicians would do well to know that there are different distributions of what they call "probabilities" out there.

I don't get it, if it's about me too. Can you clarify what you mean by "there are different distributions of what they call "probabilities"".

 
faa1947:
the ZZ forecast is from another direction: what do we forecast? level, level increase, level slope, level reversals - what do we forecast?


only what can realistically be predicted within the framework of the program used and the available input data .... there is no other way out... not what I would like... but the resulting profit will of course be small and the question arises as to the advisability of such efforts (whether they are needed... such predictions) .... i don't know how to make a good prediction from the same series(graphically ) and sometimes i myself wonder... although it's more like science fiction )))) although....

 
Vizard:


only what can realistically be predicted within the framework of the software used and the available input data...

Well, dearie me, here we go. The whole point of this thread is to define predictability. I put forward the idea that if you "properly" create a model, it will have the property of predictability. But it doesn't work, Liyuo made a mistake in modelling it, or it's rubbish at all.
 
AlexEro:
After you banned me, well, or initiated a ban for no reason, you don't exist to me.

And if you are rude without understanding, you will go back to the same place.