MoneyMakerFX Algo

• 01/15/2020 at 9:37 AM #116840
  GraHal

  Vonasi wrote:

  ALZ says it failed which should tell us all that it is over optimised/curve fitted.

  Of course it was curve fitted … it was optimised!
  It was curve fitted to a downtrend over almost the entire 100k bars that was available to me.

  Vonasi wrote:

  Optimising it on the entire 200k bars seems pretty pointless

  Not pointless in my mind as then we would have best fit / compromise values for variables over 200k bars (better than 100 k bars). More to the point … the first 100k bars (which I don’t have access to at 15 min TF) includes a long uptrend (see attached).
  So we would then have optimised values over a long uptrend and a long downtrend (last 100k bars) and we could then see if the range of values I entered in the optimiser can accommodate both down and uptrends (which I doubt as I set it up real quick to help the OP).

   

  

  GraHal

  Participant

  Topics: 222

  Replies: 9294

  Been thanked: 2461 times
  01/15/2020 at 11:09 AM #116848
  Francesco

  Excellent considerations GraHal, but i think this system is immature.
  Even if the results (on 100k) are pretty goood, the number of trades is too low, and i don’t like this.

  Anyway that was my first coding experience, so better than nothing 😀

  1 user thanked author for this post.

  GraHal

  

  Francesco

  Participant

  Topics: 33

  Replies: 440

  Been thanked: 200 times
  01/15/2020 at 12:22 PM #116861
  GraHal

  I’ve set the HenFrcFxEntryPt EURUSD 15Mv1 going on Demo Forward Test so I’ll update this Thread with performance from time to time.

  I don’t spend an inordinate amount of time on robustness and Walk Forward etc as it can be more time effective and even more exciting to get a System running on Forward Test.

  The Backtest performance stats do need to be good though else Forward Test can be a time-waster.

  I’ve had Systems that looked great on this that and the other Test  and flopped in Forward Test.

  Conversely, I’ve knocked up quick and dirty Systems that just keep on and on gradually making more than they lose!

   

  1 user thanked author for this post.

  Francesco

  

  GraHal

  Participant

  Topics: 222

  Replies: 9294

  Been thanked: 2461 times
  01/15/2020 at 12:44 PM #116866
  Francesco

  That sounds good!

  I have not a lot of experience with automatic systems, but i’m trying to get in it.
  As a statistics and economics student probably i’m too much influenced from backtests, but i never forget that markets changes, and systems also with them.
  For this reason i trust you 100% on the results variability of the WF.

  But as i said i think the ratio speed/results is not so good and don’t think this feature of the system can change so much…

  Right now i have a 6 algo portfolio, and the best of it makes 100 trades every 6 months, so i’m trying to search or develope something less “boring”… i don’t know how much time i would like to spent on cleaning the dirt.

  For this reason, i have to tell you (not with my perfect english :D) that i highly appreciate your “didactic duty”.
  Thank you so much!

   

  1 user thanked author for this post.

  GraHal

  

  Francesco

  Participant

  Topics: 33

  Replies: 440

  Been thanked: 200 times
  05/11/2020 at 2:53 AM #131047
  deletedaccount100622

  Hi @grahal how did the demo testing of this code go?

  

  deletedaccount100622

  Participant

  Topics: 13

  Replies: 116

  Been thanked: 60 times
• Author
  Posts