Designing an Automated Trading Program

Discussion in 'Journals' started by expiated, Dec 16, 2018.

  1. They

    They

    Your system is basically a MA crossover that uses a 3rd MA as input for the exit. It should take 5 to 15 minutes to code.
     
    #111     Feb 12, 2019
  2. expiated

    expiated

    It appears I will have to wait two days before I find out whether or not I'm going to get my money back...
    upload_2019-2-12_11-43-45.png
     
    #112     Feb 12, 2019
  3. expiated

    expiated

    The coder says he's going to go ahead and create the EA at the price quoted as his standard package, which is what I ordered, so hopefully I will have it available within the next three days so I can go ahead and see what it does when I back test it.
     
    #113     Feb 13, 2019
  4. expiated

    expiated



    I hope that I’m entering the final stages of designing an automated strategy that will take over the trading I currently do manually without sacrificing my success rate too greatly. The first snag I’ve hit is that the MetaTrader 4 (MT4) strategy tester seems extremely temperamental. Even though I’m doing the exact same thing every time, sometimes it works, but a lot of the time it doesn’t.

    My second challenge is that an expert advisor (EA) is, by its very nature, a relatively dumb animal. Hence, getting it to deal with “chop” efficiently and effectively is a significant obstacle that must be overcome.

    The last problem I’m having is that sometimes the MT4 strategy tester does not execute winning trades, even though all the criteria for entering a particular position had been met. This is particularly troubling because it makes me wonder if the same thing happens during live, real-time trading.
     
    #114     Feb 16, 2019
  5. Cheesy

    Cheesy


    I run a live money strategy, the same exact strategy on a Simulator and do a back test daily and all 3 are rarely the same. Getting my strategy to take every trade has been a challenge. There are things that happen in the back tester that you cannot see a reason too. All of my biggest challenges were solved by watching every trade on the simulator closely.

    It's like watching oil based paint dry, especially if it's slow and market setups are not happening.
     
    #115     Feb 16, 2019
    expiated likes this.
  6. expiated

    expiated



    In my comments following a previous test run, I mentioned that three obstacles I was having to deal with in finalizing an automated trading system involved the MT4 strategy tester not always working, it’s failing to executing some trades that it was designed to make, and configuring the expert advisor to deal efficiently and effectively with “chop.”

    Another BIG problem for me, given that my best strategies work on lower time frame charts, is that the MT4 strategy tester, more often than not, refuses to let me test my algorithms on anything below a four-hour chart.

    I therefore had to switch my focus to coming up with the best algorithm I could for the four-hour time frame, which is the one displayed in the above video clip. My plan is to take what I learned from its design and apply it to manual trading on 60-, 5-, and 1-minute charts.
     
    #116     Feb 16, 2019
  7. fan27

    fan27

    Did you figure out why the code was not taking the trades it was designed to make? Also, why would the MT4 strategy tester "refuse" for you to test on anything below a 4 hour chart?
     
    #117     Feb 16, 2019
  8. Cheesy

    Cheesy

    I have used MT4 in the past with extensive testing with experts, this problem with the 4 hour chart may be because you do not have enough lower time frame data within the sample period you are choosing. However, I am sure you have probably looked into that. Otherwise, the other culprit is just dirty Data...

    You can try using a higher timeframe to filter chop, so you would have to code the higher timeframe as a filter for the lower time frame you are trading on. Chop as everyone knows is extremely difficult to get around. The ADX indicator is also a good chop filter, it wont catch them all but it may lessen the amount of trades taking place in chop. I used it for a long time but I got rid of all indicators.
     
    #118     Feb 17, 2019
    expiated likes this.
  9. expiated

    expiated

    No, and I don't think knowing the answer will return enough benefit for me to take the time to research it, so I will probably never know unless I happen across it one day. I also do not know why the tester does nothing when I load a 60-, 30-, 15-, 5-, or 1-minute chart.
     
    #119     Feb 17, 2019
  10. Watching this second video I see that your system takes almost 300 trades to go from 10 k USD start value to about 10,240 USD end value. For a profit of 240 USD. Have you taken trade commissions into account? If not, then I fear that these will eat all (and more) of the profit.
     
    #120     Feb 17, 2019