Econometrics: State-space model forecasting - page 7

New comment
 
EconModel:

You should not scold econometrics.

By the way, for fun, it's not you, but Faa, who was scolding neural networks :))

One of the main booming areas of econometrics is nonparametric econometrics. Nonparametric econometrics is a branch of econometrics that does not require specification of the functional forms of the objects being estimated. Instead, the data itself forms the model. Nonparametric methods are becoming increasingly popular in applied research. They are more suitable for the analysis of a large amount of data with a small number of variables. Also, these methods are applied when the usual parametric specifications are not suitable for the task at hand. Nonparametric econometrics relaxes parametric assumptions, which is sometimes very useful in applied research. The main methods for building flexible models are kernel methods, spline smoothing, nearest neighbour methods, neural networks and flexible data series smoothing methods.


MetaDriver:

It's times like this when you want to hold the writer financially responsible for the bazaar. A bet, perhaps?

Tell me also that you have a (one)bar prognosticator working? Without external data?
 
MetaDriver:

At times like this, I want to hold the writer financially responsible for his bullshit. Wager, maybe?

How much are you willing to win/lose $ if I cash in this "myth" in the tester (on all instruments, for many mullions) in a day? // The timeframe, for simplicity, is also a day.

;)

If you want - use it! But bravely, without chewing the snot!

Tester, my ass!

 
FAGOTT:

Tester - fuck off

Here we go.
 

Here is the result according to the anonymous tester.

There is a parameter there: threshold - deviation of price from forecast.

For threshold = 10 pips (0.001)

threshold = 0.0005

 
TheXpert:

Tell me you also have a (one) bar prognosticator working? Without external data?
Well all my forecasters are initially one-bar. It's another thing - it's too far from 100%...))
FAGOTT:

If you want to be involved, be involved! But bravely, without chewing the snot!

Tester, my ass.

And where else would you conduct the "pure experiment" with bar colour (trading direction)? How else can you provide a 100% prediction of the direction without preliminarily reading the history?
 
FAGOTT:

If you want to be involved, be involved! But bravely, without chewing the snot!

Tester, my ass.

Let's not deviate so much from the topic.
 

I conclude from the graph: the model is not hopeless, but it won't make money.

We need to work.

The question remains: how to use the forecast?

 
EconModel:
I conclude from the graph: the model is not hopeless.
And on the horizontal axis, what? The number of trades?
 
EconModel:
From the graph I conclude: the model is not hopeless.
Imho, the graphs show otherwise )
 
tol64:
And on the horizontal axis, what? The number of trades?

time (bar number)
1234567891011121314...26
New comment