Just running some updated tests on my automated forex robot Chipper3, started with £10k balance in 2000, run the algo right up til today, Here's the 5 min data graph.
Below, run with 1 min data, a few differences but not much
On the face of it not a bad return, no withdrawals compounded in at the Very High Risk/reward settings. Of course many factors that can skew the results that are not handled by the broker and/or back testing that MetaTrader provide ie carry not accounted for, fixed spread used throughout the test, no margin changes and of course not subject to real world effects, news releases etc
As an aside there are challenges using data you get from MetaTrader through the history download feature. It seems you can only get the symbol data for standard symbol sets ie GBPUSD or EURUSD, it wont for example get data for derivations of the symbol name ie GBPUSD.sb or EURUSD.fix for example.
I therefore grab the data, say GBPUSD data for example and then copy/rename/adjust the symbol name to GBPUSD.sb and use that which is what ADS SB account requires to see in the history folder. Let me know if this is of use to you.
Would love to hear it you have a more elegant way of getting good back data to test.
What would be good of course is to have a multi currency MT4 back testing solution, in the real world I run the algo on multiple currencies which of course all have a demand on the finances, the algo tunes its order size to match balance so there are interactions across the different pairs which cant be replicated when only testing a single currency pair.
I'm using the C# to MQ4 bridge from netquotes. This allow me to code 99.99% of the algo in C#, I do have a discount code for the product if interested let me know.
I am also testing the FXPro fixed spread 1 minute data set in a similar fashion.