Ideas on Backtesting

thread: Ideas on Backtesting

  1. #1
    Junior Member pohe05's Avatar
    12

    Ideas on Backtesting

    Someone state that the 1 month of live trade is much better than 10 decades of back-test. Is that true?

    Perhaps backtest is bad and let's discuss why.

    Q1.the information you use for back test isn't same as the really market.
    A1.that's ture.But you can download the high-quility data form some internet such as dukascopy,as you exchange just 1or2 lots one time along with your slippage isn't huge with a fantastic broker,I think the back-test can 99% = the live statistics.

    Q2.the over optimization stuffs.let's me explan that this first.think about if you go back to 1 year ago and allow you to exchange this season again,will your commerce result be great?sure it'll be good,yeah?So how can an ea be bad if you optimizate it together with the information of last 1 year and back-test with exactly the same data?but as you can never optimizate with data of future,this backtest result do not mean anything.
    A2.that's ture as well.But in 1 hand,we can use backtest to check our system but not to optimizate it,in a different hand,we can utilize WFT to make decent use of back test.check this to examine detels about WFT:
    http://www.multicharts.com/walk-forward-testing/ http://www.tradestation.com/trading-...ward-optimizer

    my question:anybody know an simple way to do WFT using MT4?it takes alot of time when performed by hand.

  2. #2
    Senior Member Tataylo's Avatar
    435
    1. Someone state that 1 month of live trade is much far better than 10 years of back-test. Is that accurate?
    ....
    2. Anybody know a easy way to do WFT with MT4?it requires alot of time if done by hand.
    1. I doubt that anyone will be profitable after just 1 month of live trading!!

    Re back-testing, I agree with you that information should be accurate. I would also suggest that you need to somehow find the appropriate balance between having sufficient information to ensure statistical validity (too little information runs the risk of being'duped by randomness') and obsolescence (old information might be inaccurate because market cycles, patterns, daily ranges, fundamentals/economics and other drivers, etc have changed in the interim). Also, your underlying criteria need to be robust enough to prevent'curve fitting'.

    Much also depends upon how much of your trading methodology is'mechanical' (may be coded into an EA) and'discretionary' (principles are either too complied, or too vague, to be coded). To say the obvious, if it can not be egorized, then it can not be back-tested utilizing automation.

    If your bias is toward mechanical trading, here are two great posts which may offer insight regarding the processes required: here and here. And also this article by Merlin Jeffries. Merlin makes the great (IMO) stage that the programmer should use the information to prove a concept, rather than utilizing the information to attempt and produce a concept.

    2. Listed below are links for two simulators: vHands and Quantem. I haven't used either recently, although I did experiment using vHands temporarily back in 2008. You may prefer to give them a go. Instructions are in the linked threads.

    Hope that will help you to get started.

    David

  3. #3
    Junior Member pohe05's Avatar
    12
    1. I doubt that anyone is going to be profitable after only 1 month of live trading!!

    Re back-testing, I concur with you that data must be accurate. I would also suggest which you want to find the right balance between getting enough data to ensure statistical validity (too little data runs the risk of being'duped by randomness') and obsolescence (old data may be inaccurate simply because market cycles, patterns, daily ranges, fundamentals/economics and other drivers, etc have changed in the interim). Additionally, your underlying criteria demand...
    thanks to the articles URLs,that is really great and beneficial.
    I concur with you that NOBODY can be profitable after 1 month of live commerce.
    thanks

  4. #4
    Member Snore's Avatar
    53
    Hi,
    there's a rumor in Forex that says Back Testing is not good and it proves nothing.
    I think that this rumor started when some vendors tried to prove their merchandise via revealing some awesome back tests, but in real their customers weren't satisfied from the outcomes.
    Back testing could be so useful when you use it properly,
    First your input data have to be accurate, second choose the right version for it.
    I would suggest you to use Open Price Just and do not use optimization, instead you can use some temporarily drawing code in your EA to see what does your EA really do and then back evaluation with your eyes and find out in what situations your EA had poor outcomes, ultimately maybe you choose to bring another indior into your EA to boost its caliber. It's what I always do.

  5. #5
    2. Listed below are links for two simulators: vHands and Quantem. I have not used either recently, but I did experiment with vHands temporarily back in 2008. You might like to offer them a try. Directions are in the threads.

    Hope that will help you to get started.

    David
    Very valuable post Hanover

    Thought I would add a different simulator that might be of interest: Forex Tester

    It's a standalone program that rather looks and feels like MT4.

    I have not used it for whatever automated, though it does support such testing as well, such as this thread suggests, but I have used it to unleash market information of news events in the past to clinic manually trading/scalping high volatility post-news releases. (The trick of course would be not to examine the price history before analyzing yourself...)

    It's a good tool for those that want to acquire simulated display time on weekends and such. . Though, they've begun a subscription fee for decent tick information and won't allow you to import your tick amount data. You can just import M1 data and higher with no subscription service.

  6. #6
    I have only done WFT once on MT4, it did not seem that hard to do, but I did not see the advantage. I used it as a bit of an experiment and probably did not do it but I could not find any point in paying more effort.

    With the usual crappy historic statistics, WFT can not give you an advantage over customary back testing. Its garbage in, garbage out. I believe that the focus should be on information quality, not a fancier testing scheme.

  •