We are trading at CFH Markets using PAMM allocation on their MT4-platform which is working great. You may contact us for further information (email@example.com) .
It is critical to use tick-data in all backtests that is performed on the MT4-platform.
The data provided by Metaquotes is not complete (several missing days and even weeks in some periods) which will result in incorrect and unrealistic results. We have been in contact with Metaquotes and they will not correct or improve their data. The data is overall low quality and backtests performed on this data should not be considered accurate and absolutely not used for investment decisions. (This is the main reason why 95% fails when trading with EA on live accounts, as real data is nowhere near MT4-demo data…)
For accurate backtesting you need high-quality tick-data that gives a Model Quality of 99%, but the backtest also needs to be performed on a correct spread-setting.
MT4-demos normally uses fixed 0.9 EUR/USD spread in their backtest which is probably not(!) the spread you will be trading on live (probably you will trade on 1.5-3.0), even if you may get a spread of 0.9 it will not be fixed and this must be considered. (This is the second reason why 95% fail when trading live because they think that their scalp-system will work as good on a live account with double or triple the 0.9 spread – and it will not! Not even close …)
So to be able to backtest accurate, you need:
1) find good tick-by-tick data (contact us by e- mail, we will help you!)
2) import it to MT4 in offline mode (so Metaquotes don’t mess it up)
3) use a spread-script to increase the spread to whatever your broker is offering you when trading live
4) perform a backtest on tick-by-tick data (it will take time !) but it will result in a 99% model quality and gives you a good approximation on how a real account would have performed during the same period.
*Due to a programming mistake in MT4 it is only possible to run backtests with a maximum of 2 gigabyte fxt-file (data-file) which result that you may only be able to import tick-by-tick data for aprx 3 years (a CSV-file of aprx 3.5 gigabyte). So for longer periods you must do two different backtests.
However, it is difficult to get tick-data before 2007 so this any big problem right now.
We have written an explanation regarding the importance of correct backtest on our website, it is written in Swedish but you may access it with our translation tool (select the English-flag in top menu)http://priv.ciceron-am.com/mm/index.php/information/oevrig-information/viktig-information-angaende-backtest