Forward Trading and Backtesting difference

Hi,

I'm using a trading strategy based on 2 timeframes and when I apply it live it seems to work well. But I've created an EA and the results are quite bad. My system is very mechanical and I don't understand why the backtesting is so bad. Does the use of a second timeframe have an inpact on the backtesting process?

Cheers,
Daniel

Did you look at it manually? I mean did you look if the backtests are executed correctly and they don't miss any price action in between? I would look into that. If they are executed correctly, then you have a losing system in your hands.

I'm not an expert on Experts but are your time frames below H1, I remember having a similar problem on the shorter <H1 timeframes where each backtest came up with very different results! You might also want to refresh data in case it's corrupted.

Like Custos said, checking trades are being executed correctly seems like the place to start.

2nd timeframe shouldn't have any problem as long as you manually change this through either external int or through conditional operators.

How have you backtested? If you download automatically from MT4 server or use the same MT4 data that is imported during sessions you have very bad data. You need to download M1 data from a reliable source OR tick data (tick data is a big headache to get to work) and use a separate MT4 that is updated and then remove all connection and account data so it doesn't connect to the server or logs onto your account and delete all historical data and replace it with reliable M1 data, and then convert this M1 data to other timeframes with the script that comes with MT4. Or have you used another program? Backtesting can only be as reliable as the data you use, even $3k simulation software can't do shit with bad data.

Remember that backtesting with 1M data even on H1 timeframe is not completely reliable, below 1H it might be junk but to be honest I have no idea. Quality of data with 1M data drops as timeframes become shorter, where the cutoff lies for "minimal difference" and "too much difference" I do not know.

And what does "work well" mean for you? 50 trades and 5% ROI profit per trade means nothing. 1k trades and 25% ROI profit per trade means your system is profitable, although it could be much lower then 25% ROI (but also higher!). I think that based on win%/loss/win and using simulation software to generate like 100k instances of whatever number of trades you currently have allows you to see what the chances of your system being a profitable one in the long run are better then backtesting.

maybe measure the distance between yourself and your keyboard to get an idea.

Around a foot?

Thank you for all your answers. I think I might have bad data. But also I don't understand why we have to get good data, because in backtest I have the same data that I've received during live trading. This is really something from metatrader that I don't understand.