What would be interesting to learn with Machine Learning of Algos ?

Forums ProRealTime English forum ProOrder support What would be interesting to learn with Machine Learning of Algos ?

Viewing 8 posts - 1 through 8 (of 8 total)
  • #129510

    Hi Huys,

    If Machine Learning is the Science of Learning the Good and the Bad of past trades, what do you think would be interesting to Learn ?

    May be just a stupid idea but one think who will be interesting will be the previous highest/lowest high/low as invalidations of trade (previous donchian so far). The pupose will be to diminish the Stoploss for the next trade and as each trade as a constant price to rise the % of win trades.

    Just an idea I think and will code next days. With Array function it should be easy

    If you have some others ideas of what would be interesting to know (difference between 2 average, previous angle of average as in Vectorial Dax …)

    See U

    1 user thanked author for this post.
    #129537

    Hello zilliq,

    I’m surprised so many people are drawn to machine learning here on the forum. In my opinion if you are not already capable of designing robust and profitable strategies without ML, it’s foolish to think that ML will help you. Besides PRT is not really design to handle such complicated science. The problem is that I don’t see a lot of strategies on the forum that makes sense at their core, because people have not traded manually before and so don’t really understand how markets work. To answer your question, previous donchian is a good start, what works better is a a fixed percentage below /above the close that breaks the donchian to avoid wick

    #129538

    Agree with you, but we all put inside “Machine Learning”

    I don’t really believe in Machine Learning but I thinks that as people we can learn from our mistakes 😉

    Thats’s why some thinks should be correct by previous lost/mistakes

    The main problem for me is the slow capicity of PRT to calculate. I add many code in my algos I should delete because the backtest take many many hours..

    The second problem is that the past is the past and nothing sure it will be replicate in the future

    And the last but not the least many lost trades are totally unpredictable so even with “ML” it will be difficult to no replicate the same mistake

    See U

    #129540

    @stefou102 Don’t agree at all, I think a lot of people in the forum wish didn’t went through manual trading and lose a lot of money before getting here. You can learn and understand the market also by studying hard, as you can create algo systems regardless of the work that a person can do manually and going beyond the classic manual trading conception.

    #129542

    There are as many traders’ profiles as there are traders.

    #129546

    Hey Fransceco,

    agreed if you know where to look to learn. In any case you lose money either on the market, or by buying lot of outdated books, or by buying expensive training. Some ways are cheaper than other;)

    #129565

    The most interest is convert Vectorials in machine learning.

    #129579

    Yes but Fran55, we’re all still waiting for you to post your improved DAX Vect code so we can get started on the ML version 😉

    4 users thanked author for this post.
Viewing 8 posts - 1 through 8 (of 8 total)

Create your free account now and post your request to benefit from the help of the community
Register or Login