Not logged in
Viewing 6 posts - 1 through 6 (of 6 total)
  • #237880 |
    Customer
    48 Posts

    OK. I’m using SQX Build 116.76 and Dukascopy M1 data from 2003.5.5 – 2015.12.31 for my Random build of some new H1 strategies.

    I’m trading a fixed amount of $100 per trade.

    The only filters I’m using are:

    1. # of trades > 500

    2. Profit factor >= 1.3

    3. Max DD% < 30%

     

    I have tried almost every building block, preset configurations (i.e market, trend following, mean reversal, Fuzzy), and combinations thereof with very simple stop entries and almost every type of exit. SQ has produced over 5 million strategies but only 2 of them pass those basic filters but both have a very low net profit. This seems highly unusual.

     

    Are others seeing similar results and, if not, what do you do to generate viable strategies for USDJPY?

    #237883
    Customer
    18 Posts

    Hi afhampton, I just tried using all of your parameters including date range, random generation, stop entries etc and got 6 strategies that passed the filters in about 10 minutes

    Over 60% are being rejected by the profit factor filter and # of trades filters, the rest mostly dismissed by the automatic filters

    The rest of my config will obviously be different, if you want to post the config or PM it I can run it again using your exact settings if you want a proper like for like comparison

    #237885
    Customer
    48 Posts

    Very interesting Ash24FX.  I would love to see your build config and how you got that kind of response. I have attached one of the many configs that I have tried to see what you come up with. Thanks for the help!

    Attachments:
    You must be logged in to view attached files.
    #237891
    Customer
    18 Posts

    Hi afhampton, only managed  to generate 1 strategy in about an hour with your build settings. I’ve attached the build config I was using for comparison, some of the settings I’ve tweaked to match your build settings like Fuzzy Logic, Exit on Friday etc. Generated 15 strategies in less than 10 mins using these. Not saying they are all great strategies by any means, but they all pass the filters. At least it gives you something to compare against.

    Ash

    Attachments:
    You must be logged in to view attached files.
    #237904
    Customer
    48 Posts

    Thanks for the help Ash24FX. That’s fascinating to me that you were able to generate 15 strategies in 10 mins. I’ve been running the same build for 6 hours 49 mins on a server with 32 virtual cores, 128 GB of RAM and it is the only thing running. It generates over 300,000 strategies per hour yet I have zero that have made it past those basic filters. I’m going to try it using your config. I noticed that you have it set with 0 slippage, no commission, no profit target required. Can you share your reasons for leaving those blank?

    #237907
    Customer
    18 Posts

    The PC I’m using has 16 cores (32 threads) and 32GB RAM, but probably only using half the RAM at most when generating strategies. SQX is still quite new and I’m still learning all of the new features so I like to keep my build settings fairly simple for now at least to generate the widest selection of strategies I can. I do use quite strict filters at build stage but can still end up with a few hundred strategies per day depending on the pair, timeframe etc, I then filter even more aggressively with the retester using various robustness tests. I leave slippage and commission until then as well so that I can test using different values to see how it affects the strategies, I just find it more flexible that way, personal preference I guess. The only Exit Type I currently have set as required is stop loss, I have generated strategies without a stop loss and they can look really impressive, even in backtesting, but I would never trade them on a live account. I leave all of the other exit types selected so they can be used, but don’t have them set as required, that may change in the future but for now while I’m still learning it allows me to generate the widest range of strategies possible to see what works and what doesn’t

Viewing 6 posts - 1 through 6 (of 6 total)

You must be logged in to reply to this topic.