Econometrics: one step ahead forecast - page 8
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
Fuck it!
That is what the article says. It is suggested that we collectively refine the model or at least figure out what is wrong.
? (how to help?)
? (how can I help?)
Read the topic from the beginning. The model that produced the forecast for tomorrow is now in effect. Is it a good model, or does it need improvement?
Read the topic from the beginning. The model that produced the forecast for tomorrow is now in effect. Is it a good model, or does it need improvement?
I will read the articles at the weekend (no time to mow now:))
Let's read the topic from the beginning. The model that was used for tomorrow's forecast is now in effect. Is it a good model or something needs to be improved?
I don't understand why shout at every turn that econometrics is "the best" and at the same time ask for a model to be developed? )))
the whole thing reminds me of Yusuf's scenario... an article, then a request to teach it...
about the forecast where the lags were removed... all this was done long ago, not manually... but with the help of NS - first comes the automatic selection of inputs... their optimization and output of the model...
the same econometricians use... p-square may be 1 and other errors may be 0.999999999999999 and better... but this will not make it better... in all cases (with 1 VP) this is just a simple fit to the history... no more...
And for a forecast with an error of 50-100 points such fuss is unnecessary...
I don't understand why shout at every turn that econometrics is "the best" and at the same time ask for a model to be developed.
What's not clear? There are an infinite number of models. I hope that in the process of collective development the word "ze best" will become concrete.
I love reading threads like this. It's like watching UFO movies for me. I don't believe in UFOs myself, but the topic is terribly interesting.
Here is the formula that was used to make the prediction:
kotir hp1(-1 to -4) hp1_d(-1) hp1_d(-2)
It takes the last four values of indicator HP - no problem in the EA. But it takes two last values of residue between the indicator and kotir. And it turns out that it is necessary to take two, not four. At the same time, we know that this complex indicator reflects the movement of the kotir at 97% (R-square value). Take a single indicator or a whole bunch of indicators and answer the question: how much does this pile correspond to the quotes? Here is an automatic result instead of wandering in the dark.
Make the effort to write a simple formula and you'll realise that it's not UFOs, it's reality
Did it already. Look in the codebase. My advice to you, understand the mathematics of econometric models, read the history - say, who first derived these formulas and with what assumption. Then you will understand everything. Many people have an irresistible urge to put the last N bars of history into a black box, a crystal, and it will magically give us a prediction of the future. And the wiser the formulas in the box, the more faith in their result (like Yusuf - he is my idol).