From theory to practice - page 518

New comment
 
Renat Akhtyamov:

no more than 10 minutes or until a new sufficiently risky transaction in the market

if the latter is not available, recalculate

It depends on the timeframe, which has its own trend sections.

 
Novaja:
OK, taking the last point, i.e. we know the state of the system at this point, how long will the state of the system be stable in the future to be able to predict?

it is always possible to predict, as one state transitions to another and a prediction can be made according to this.

 
Novaja:
Victor has an example of an EMA-based passback filter in his kodobase,
https://www.mql5.com/ru/code/192

What he writes on the subject:
The smoothing result will be the same as using a zero-delay filter (symmetrical impulse response) except for the edges of the sequence where the edge effect, or as it is called here, overshoot will occur. MA, i.e. filter with finite impulse response, was used above as an example. If you use filters with infinite impulse response (EMA for example), theoretically edge effects will extend throughout the whole length of the sequence.

Redrawing by logic is more of a boon than an evil, as it allows systematization of states that are interfered by noise components, which are states, i.e. useful information on a smaller timeframe...

 
Andrei:

Rewriting is logically more of a benefit than an evil, as it allows the systematisation of states that are hindered by noise components, which are states, i.e. useful information on a smaller timeframe...

Andrey, you're a genius, I really missed it, in a good way))
 
In general, the formula D = Sqrt(c * lambda * t) is clearly missing something else. Inertia or acceleration.
 
Novaja:
Andrei, you're a genius, I really missed it, in a good way))
No, it's been explained here a hundred times by different people. ))
 
Andrei:

predictions can always be made as one state transitions to another and predictions can be made accordingly.

completely coo-coo?

 
Maxim Dmitrievsky:

Are you totally kooky?

Are you trying to start a squabble here? I'm not interested...

 
Andrei:

Do you want to start a squabble here? I'm not interested...

No, I'm calling for sobriety.)

 
Smokchi Struck:
x@@@@@vo! )))

figure out how to improve it.

Well, that's what I assumed ;)))

1) polynomial regression is applicable for approximating fixed (not varying) data (polynomial order of 5 or less). The model can be used to interpolate intermediate values. But it is not applicable for extrapolation beyond the approximation interval.

2) polynomial regression is a very bad idea for approximating dynamic (changing) data.

1...511512513514515516517518519520521522523524525...1981
New comment
Reason: