prove why neural network will not work in trading

Discussion in 'Automated Trading' started by trend2009, Jul 10, 2019.

  1. Overnight

    Overnight

    Yes sir.

    I just keep trying to fly by the seat of my pants. The hell else I have to live for?

     
    #31     Jul 12, 2019
  2. As long as you have risk limits in place and know when to bail out when being wrong. And as long as you have done your math that the target levels you are aiming for outperform the sum of failed attempts in aggregate (speaking of course in terms of a larger picture not this one single particular trade only). I trust you have done so if you are in this business to generate positive returns.

     
    Last edited: Jul 12, 2019
    #32     Jul 12, 2019
  3. IAS_LLC

    IAS_LLC

    Gated recurrent unit long short term memory recurrent neural network....

    I played with LSTMs for a minute....I don't remember why exactly I quit them, but they weren't working for what I tried them on
     
    #33     Jul 12, 2019
  4. If anyone is able to develop a profitable system based on neural network technology, assuming you are going after gaming the Emini market, the next hurdle is getting it profitable in light of the huge CME exchange fees.
    Currently, I am paying $1.30 per contract. This is a criminal level. For high frequency trading, that must be lowered to $0.25 or lower. To do that, you must become a member of the CME.
     
    #34     Jul 15, 2019