Log in

Extreme extrapolation with extrapolator

Here I want to show you an extreme case of the use of the Extrapolator. In the data mining it is a common aprroach to combine the techniques. 

Here I do what I consider the right way to use all the excellent extrapolators. They are in fact ment to be used together and to be combined.

The yellow line is the neural net outut from the BPNN with Hodrick -Prescott filter (a mod of the BPNN)

The blue Line is the indicator Extrapolator (mql code base)

The pale green is the extrapolator Sinc-MA (mql code base)

As you can see those lines are giving intervals where the price is expected to go. It is unvise to take only one extrapolation, it is wise to have many.

 

image
От Поле за пускане

 

Comments

  • JohnLast 3512 days ago

    I think this is the way to use the availabe extrapolator indicators, forming a commitee from them.

     

  • jaguar1637 3512 days ago

    Hi

    There are several types of Extrapolator indicators.

    Their reliability is based on type  of algorythm, number of bars analyzed.. but for sure, the timeseries as input values doesn not reflect how is the market. Because, the market is like a torrent or a river.

    There are several factors to take into accounts, outside the vision of the clouds created by the orders :

    - the volume of orders (more there are many orders, the markets goes throw the direction given by the biggest orders.)

    - speed regarding how the market is reacting.

    - the current oscillation between ask and bid, sometimes, there is a cross between bid and ask

     Those extrapolators does not take those parmeters into account. I tried for 2 years many extrapolators and their results are less than expected at first glance

    I thought using simultaneously a bunch of extrapolator (fourrier/ Burg/ Gauss) about fetching results. But I was not statisfied