Econometrics: one step ahead forecast - page 125

New comment
 
Colleagues, in my haste, I called it wrong, not "envil" of course, but EViews. Sorry. I was too lazy to go back and look up its name, so I blurted it out :o)
 
Mathemat:
Where did this idea come from? Why were you so sure about it?

You are fixated on the quotient alone. You can differentiate it (take the differences) as many times as you like, even 10 times. But what good is it, where is any guarantee that it will continue to be the same in the future? Where are these guarantees in the model itself (along with their estimates)?

Let's kill the non-stationarity. At the end, we model a whole range of variability of variance with GARCH
.

Stability can be found in other price functions, not only in the quotient itself. To do it you have to strain your brain and stop chasing quotes in only one way (by regressions of charts).

Agreed. Theoretically know more about periodicity (not to be confused with seasonality). Let's make some suggestions.

Where did this idea come from? Why were you so sure about it?

That's the idea. Maybe it's not neatly implemented, maybe it's just nonsense.

 
faa1947:
I say let's identify the different characteristics of kotir, the most unpleasant among them is non-stationarity.

http://imglink.ru/pictures/10-01-12/4396b8f3b63f2887d8db33cea9380a09.jpg

You are looking for close - the green line of the indicator, while I need blue and red lines for trading. At first glance my lines are much more stable than yours, I have not applied any mathematical transformation, the indicator consists of 10 lines. I think you are looking at the market from the wrong "angle" - there is no future, only the recurrence of events

 
Farnsworth:

(1) Take a fixed sample length for which you are sure that EViews identifies the model correctly.

(2) Sweep the sliding window in the direction of "astronomical" time in steps of one bar (counting)

(3) For each such step (let there be at least 100, taking into account manual work) let EViews determine the optimal coefficients of the selected model

(4) Write it all down in a table: step number, coefficient values, error/residual, criteria

And look at it all together, how it all changes.

The results that are posted in this thread above are obtained this way. The profit factor is just above 1. I have come to the conclusion that the model has no predictability and I'm stuck with that. For smoothing, HP was applied with ladd = 1. Could be here. But it's not clear what "predictability" is. If you look at what you get in the tester, the model doesn't hold the trend and it's not about false reversals.
 
Avals:
faa1947, price increments depend on a huge number of factors, some of which are not even related to previous prices. Whichever method you apply, even in an ideal case it only takes into account a small fraction of them. Most of the time, their influence on the quote is insignificant - at the level of noise - and quite rarely they come to the forefront and are able to create some movement. Then there is an opportunity to match them using certain cause-effect relationships. So, don't initially build a model that is always trying to predict something - filter the bazaar. And build subject-oriented models, not just what's in your software. What processes may be behind pricing based on the basics of trading. Speculative, investment, conversion, etc. Certain assumptions about the mass behaviour of traders, building a model on this basis, verification (maybe even based on ecometrics).
No doubt. Called state space. There was a handyman here, made some promises and no promises. Ready to collaborate on this topic with everyone. So far the only state space I know of is the dollar index. Attempts to include stock exchange indices have failed.
 
IgorM:

http://imglink.ru/pictures/10-01-12/4396b8f3b63f2887d8db33cea9380a09.jpg

You are looking for close - the green line of the indicator, while I need blue and red lines for trading. At first glance my lines are much more stable than yours, I have not applied any mathematical transformation, the indicator consists of 10 lines. I think you are looking at the market from the "wrong angle" - there is no future, if only the repetition of events

I have indicators that are better than yours. The regression I will apply in this thread gives an in-sample profit factor over 50. But we are trading outside the sample, hence the conversion.
 
faa1947:
No doubt about it. It's called state space. There was a handyman here, made some promises and no promises. Ready to collaborate on this topic with everyone. So far the only state space I know of is the dollar index. Attempts to include stock exchange indices have failed.


The most important thing is the correct formulation of the problem, and the formalisation of the solution is secondary.

 
faa1947:
I have indicators which are better than yours. The regression that I will apply in this thread gives an in-sample profit factor over 50. But we trade outside the sample, hence the conversion.

http://imglink.ru/pictures/10-01-12/c556d9f9b5fbe778a5b07fed36d8ab88.jpg

so my indicators are worse than yours, but they show market conditions rather than smoothing, differentiating and ....

It is important to formalise the state of the market, and once this task has been completed, you can move on to forecasting.

 
Avals:


the most important thing is the correct formulation of the problem, and the formalisation of the solution is secondary.

The state space is extremely attractive. But it is probably more applicable to the stock market, to macroeconomics.
 
IgorM:

http://imglink.ru/pictures/10-01-12/c556d9f9b5fbe778a5b07fed36d8ab88.jpg

so my indicators are worse than yours, but they show market conditions rather than smoothing, differentiating and ....

It is important to formalize the market condition, and once this task is accomplished, you can move on to forecasting.

In my smoothing and differentiation there is quite a definite idea and purpose.

Formalisation cannot be the goal - very quickly you get a game of numbers that are confirmed by the tester and then surprised. The most beautiful figure is ZZ.

1...118119120121122123124125126127128129130131132...139
New comment
Reason: