Hello fellows.

Which one is the most "obscene" picking into the future oscillator?

I want to experiment with creating an NN indicator to predict probabilities into P.I.T.F. indicators.
0
438
4 Replies

Bookmark

Sort
I think you mean "peeking". I'll change the title.
Picking is the same as "choosing" (something).
1
Take a ROC and shift it one bar into the future.
1
QUOTE:
Take a ROC and shift it one bar into the future.

This is an interesting approach I am working on. Also some of the Indexlab indicators are interesting, despite not flagged as PITF.
0
QUOTE:
is the most "obscene" peeking into the future

It depends on what you mean by peeking. If you're trying to extrapolate into the future, then look at the TSF indicator. It performs a first-degree linear regression for its extrapolation.

If you want the coefficients of the regression to be partial derivatives (basically a partial differential equation(s)), then you should look at fintantic's Kalman filter. And that particular implementation will take two inputs so you can have a price action indicator and a volume action indicator as inputs. https://www.wealth-lab.com/extension/detail/finantic.Kalman

In theory, a system of partial differential equations will give you the best dynamic solution, and that's why engineers commonly employ the Kalman filter in all high-end flying guidance control.

You mention the term "oscillator". If you were hoping to solve this problem with coefficients in the frequency domain (Fourier coefficients), forget it. There's not enough frequency energy to get stable Fourier coefficients from price and volume action.
1