Not logged in

Forums

Viewing 15 posts - 1 through 15 (of 19 total)
  • #116702 |
    Participant
    12 Posts

    There is a new product on the market called EA Controller  (http://exceptionalfx.com/ea-controller/ )  which essentially only allows copying from demo to live account while the EA is in an uptrend.

    In order to back test to see if this improves performance on live, we need to be able to do a WhatIf analysis on the results from back testing the EA in which a MA (std features such as type, “period” (= number of closed trades), and Shift, is applied to the balance curve. Then the WhatIf analysis would analyse the back test history for only those periods when the Balance is > MA of Balance, ie in uptrend. That is, discarding all the results for when Balance was < MA of Balance, ie in downtrend/DD.

     

    The attached image shows an example of a MA of Balance applied to a Balance Curve. The trades within the green boxes would be the ones captured in the WhatIf analysis.

     

     

     

    Attachments in this forum are visible only for registered users.
    #142780
    Moderator
    2349 Posts

    Hello,

     

    if I understand correctly, you can do this type of analysis in QA using the Equity control feature. 

    #142782
    Participant
    12 Posts

    Hello,

     

    if I understand correctly, you can do this type of analysis in QA using the Equity control feature. 

    Thanks. That is what I need. I assume it is a SMA?  Any chance of offering a SMMA function?

    #142800
    Moderator
    2349 Posts

    Hello,

     

    most traders use this type of MA. Do you see any advantage using some other type?

    #142808
    Participant
    12 Posts

    Hello,

     

    most traders use this type of MA. Do you see any advantage using some other type?

     

    As pointed out in the explanatory article, equity control does not work well when the equity curve is choppy. And when the equity curve is choppy, the SMA also tends to be choppy.

     

    So if we use a SMMA rather than the more closely following (more choppy) SMA, we should be able to better handle a choppy equity curve. That is, not get whipsawed so much. At least that is the theory.

    #143013
    Moderator
    2349 Posts

    Hello,

     

    ok, I will check this if SMMA could be added by creating new snippet code

    #143019
    Participant
    12 Posts

    Hello,

     

    ok, I will check this if SMMA could be added by creating new snippet code

    Thanks. That would be great.

     

    Even better would be if you could also add a Shift function for the MA.

    #143037
    Customer
    324 Posts

    Well this might work if you can load 10 years of history trades in MT4 and create the moving average based on that. Then you can stay out of stagnation periods that might start the day after you launch your strategy .Would be smarter to be able to block trades until a certain amount of consecutive loosers occoured wich could be in line with backtest and first then allow the strategy to trade.

    #143062
    Participant
    12 Posts

    Well this might work if you can load 10 years of history trades in MT4 and create the moving average based on that. Then you can stay out of stagnation periods that might start the day after you launch your strategy .Would be smarter to be able to block trades until a certain amount of consecutive loosers occoured wich could be in line with backtest and first then allow the strategy to trade.

    Worth considering. The thing is you also need to have a decision point for when to stop trading again. After x wins?

     

    A z score decision is also worth considering; by itself or in conjunction with an MA.

     

    However you certainly don’t need 10 yrs of data. In tests so far, history of several hundred trades would be sufficient.

    #143063
    Customer
    60 Posts

    Worth considering. The thing is you also need to have a decision point for when to stop trading again. After x wins?

     

    A z score decision is also worth considering; by itself or in conjunction with an MA.

     

    However you certainly don’t need 10 yrs of data. In tests so far, history of several hundred trades would be sufficient.

     

    I think the interest is not to start and stop trading only, but also change position sizing.
    ‘‹The idea would be to trim the trades which have the highest probability of failing. For example: extreme X% of consecutive loss/win sequence.
    ‘‹Would need to be linked to a databank where we input for each magic number the threshold of consecutive win/losses as per its historical distribution.
    ‘‹Getting complicated… But if it is just an MA aren’t you worried of entering after the battle is won already?

    La PIP c'est chic, 😀

    #143065
    Participant
    12 Posts

    I think the interest is not to start and stop trading only, but also change position sizing.
    ‘‹The idea would be to trim the trades which have the highest probability of failing. For example: extreme X% of consecutive loss/win sequence.
    ‘‹Would need to be linked to a databank where we input for each magic number the threshold of consecutive win/losses as per its historical distribution.
    ‘‹Getting complicated… But if it is just an MA aren’t you worried of entering after the battle is won already?

     

    Lagging from MAs can of course be an issue as it is in price charts, but the example i want to show is for an EA where the use of MA 30 almost doubles the return and halves the DD, and hence where lagging is not an issue.

     

    https://www.screencast.com/t/X0yBU4tkT

    #143066
    Customer
    324 Posts

    yes i did some test in QA on strategies that have been really profitable in the last 6 months of live trading on real account. They all performed terrible no matter what Ma i selected. I also tested it on some loosers and there it lowered the drawdown but they were still loosers. So that means that you could use it on loosing strategies but not winning strategies.. so which one is the winning next month and which one is loosing.. hehe..   The best way would probably be like i said to load as much history trade data as possible and apply a 200MA to make the decision to take a trade or not. 

     

    Now!  Lucca has a good solution and it is already used by some traders using their home made excel sheets with live data from Fx blue connected and they seems to be doing great proven by 100% value increase in portfolio value during a year. But you would still need to have a profitable portfolio and then use it to try and improve it.

    #143116
    Customer
    324 Posts

    If the EA controller would work on a portfolio of strategies ( as one) and/or you could load historical trades i would definately buy it. 

    #143123
    Participant
    12 Posts

    If the EA controller would work on a portfolio of strategies ( as one) and/or you could load historical trades i would definately buy it. 

    Almost but not quite what you need I think.

    You can set the Controller to monitor and control a portfolio up to 10 magic numbers/EAs but at this stage, they must be for the same currency pair.

    But, in the version to be released tomorrow (v03.02), it does read however much historical data you have in the MT4 History for the magic number(s) selected.

    #143158
    Customer
    60 Posts

    If the EA controller would work on a portfolio of strategies ( as one) and/or you could load historical trades i would definately buy it. 

    Hi Sydap:

    ‘‹- copy or not copy trades from magic number when X condition is met
    ‘‹- X condition to be unique per magic number
    ‘‹- And/or change the position sizing (lots size) depending on portfolio (account) metrics
    – X conditions being for example, profit factor, consecutive win/loss, DD, negative profit
    ‘‹

    A portfolio is made of different currencies and time frame, all EAs have a unique magic number.

    ‘‹Is that in your plans?
     

    La PIP c'est chic, 😀

Viewing 15 posts - 1 through 15 (of 19 total)

You must be logged in to reply to this topic.