Not logged in
Viewing 5 posts - 1 through 5 (of 5 total)
  • #113630 |
    Customer
    484 Posts

    I’m going to be starting a series of threads in the next few days or couple weeks mostly educational and inquisitive. I’m making so videos on youtube as well.

    This is the second thread of that series. The first was already posted on adjusting your generating settings for optimal results and speed.

    Are you using and testing on tick data(SQ)? Why?

    Ask yourself-
    Is the strategy trading on M1(which benefits from tick data)?
    Is it trading tick charts?
    Is it HFT (high frequency trading)?
    M5 strategies can also fall somewhere in the “grey zone”. They don’t need tick data but it could be helpful for some M5 strategies that scalp very quickly.

    If its not on those timeframes there is no need for tick data on m15 M30 H1 H4 D1.

    This also may help people using special indicators like “renko bars”. To use them on Duka data you must export it as M1 I believe since the indi is designed to convert MT4 data.

    #129943
    Customer
    557 Posts

    Hi Theshold,

     

    Interesting topic and I’m wondering this myself.  I generate and use strategies M5 up to D1. For M5 strategies it depends what the average trade size is, for my EURUSD strategies it is around 40 pips.  I don’t think that would be considered scalping and backtesting M1 or tick data doesn’t make much difference to the results.

     

    I am interested in tick data backtesting in MetaTrader4, and am currently looking at TickStory and Birt’s TickDataSuite.  TickStory works, is free and is easy to use, but it does not solve the FXT file size issue with MT4, meaning backtests cannot run for more than a few years at a time.  MT4 really sucks.

     

    If anyone has experience of using tick data (rather than M1) to generate renko and range bars please let me know.

     

    Cheers,

     

    Mike

    #129973
    Customer
    484 Posts

    MT4 backtest isn’t based on length, its based on # of trades I believe.

    #130041
    Customer
    432 Posts

    My way of generating is to always use “selected timeframe only” mode. Then, retesting, but with tick-simulation and kick out the ones that deviate to much in terms of net profit / dd to the selected timeframe only mode. This kicks out all the strategies already that relay on intra-bar exits and small movements and leaves the one which are pretty much not sensitive to intrabar moves. These ones also translate perfectly to the backtests in live trading if backtesting the same timeframe used during live trading. The entries and exits even matched down to the pip so far (traded via MT4).

    #130052
    Customer
    484 Posts

    In my experience as well. I think people having poor experiences with live trading did 2 things:
    Over optimized
    and used too short of data.

    I see and hear of alot of people using just 2 or 4 years of data with their strategies. I tend to believe the strategy will last 1/2 as long as the data used to make it, or less.
    So strategy made on 2 years of data might last 1 year or less, if they are lucky.
    They also are trying to get rich quick which means they gona end up poor fast, robust strategies also risk less and look to survive in the market for the long term.

Viewing 5 posts - 1 through 5 (of 5 total)

You must be logged in to reply to this topic.