Index futures automation

Discussion in 'Journals' started by hilmy83, Jul 3, 2022.

Can a fully automated trading strategy work in the long run?

  1. YES!

    56 vote(s)
    66.7%
  2. Hell naw.

    15 vote(s)
    17.9%
  3. I don't know, I got my own trading to worry about.

    13 vote(s)
    15.5%
  1. hilmy83

    hilmy83

    Backtest MES from 3/14/22-11/4/22 (spring DST)
    entry - random
    single position
    stop -$50
    exit- time based end of day

    This is to simulate "trend trading". The truest adherence to cutting your losses short and let your winners run (in context of daytrading). Has +EV, but having to endure weeks or months of losing would be almost unbearable mentally (20% win rate!)

    upload_2023-1-14_21-37-23.png

    upload_2023-1-14_21-37-39.png
     
    Last edited: Jan 14, 2023
    #171     Jan 14, 2023
    shuraver and Overnight like this.
  2. Overnight

    Overnight

    Believe it or not, those stats, and especially that final chart of balance vs open equity, is very encouraging. Good work.
     
    #172     Jan 14, 2023
  3. hilmy83

    hilmy83

    Same test but adding scaling into position

    entry - random
    5 positions scale in max
    stop -$300
    exit- time based end of day

    Doubled the win rate to about 40%
    upload_2023-1-14_22-26-0.png

    upload_2023-1-14_22-26-24.png
     
    #173     Jan 14, 2023
  4. Overnight

    Overnight

    Again, equally encouraging with the scale-up in size. It's working, man.
     
    #174     Jan 14, 2023
  5. hilmy83

    hilmy83

    This is the same test:
    except the exits are all end of day. Basically a "no stop", "no target" intraday

    upload_2023-1-15_0-59-43.png

    upload_2023-1-15_1-0-4.png

    So in a true 50/50 price movement, I should theoretically would be b/e since there would be equal distribution of wins and losses even on a time based exit.

    But it's the way I enter that initial position is what's significant.

    ES/MES is naturally a mean reverting market, but it will usually make it's daily high or low within short time of the Open.
     
    #175     Jan 15, 2023
  6. NoahA

    NoahA

    Out of curiosity, what do you mean when you say the entry is random but you're trading with the trend? You mean you first identify a bull market and then enter randomly long? Conversely, if your algo identifies a bear market, it will only short but at random times?

    Or is it simply a random entry, long or short, and then it either stops out or the trade keeps going if it keeps increasing in profit?
     
    #176     Jan 15, 2023
  7. hilmy83

    hilmy83

    Random entry just basically using a breakout (whichever side is hit first), then leaving the trade open until close.
     
    #177     Jan 15, 2023
    NoahA likes this.
  8. hilmy83

    hilmy83

    same test but on the MNQ

    50% win rate like MES, but still has +EV...

    Next tests to do what happens when you scale in

    upload_2023-1-15_7-20-55.png

    upload_2023-1-15_7-21-14.png
     
    #178     Jan 15, 2023
  9. hilmy83

    hilmy83

    Same test but with b/e stop added in. 90% win rate, but because no stop, big losses overcome the many mini +1s

    upload_2023-1-15_8-20-53.png

    upload_2023-1-15_8-21-20.png

    Those large impulses are basically trend days.

    You could get frequent wins and get screwed on those days...or you try to capture the big days and try to minimze the many small losses.
     
    Last edited: Jan 15, 2023
    #179     Jan 15, 2023
  10. hilmy83

    hilmy83


    I'm going into a backtest rabbit hole lol

    Same test but setting max loss


    PF improved. Same high % wins, but of course you sacrifice your R:R. But still +EV

    upload_2023-1-15_8-43-43.png

    upload_2023-1-15_8-43-59.png
     
    Last edited: Jan 15, 2023
    #180     Jan 15, 2023