Differences in backtest on data from ForexTester and Alpari

Hello,

I have tested my system in two ways - one is demo account on Noble Markets, historical data from ForexTester.com and results are very good (for both Every Tick and Open Prices Only, on data from many years). I have tested the same system on demo from Alpari, data downloaded with Download button in History Centre (because for Alpari, MT4 does not download data from MetaQuotes, but its own) and results are much worse.

In both tests I have used the same spread, the same pair and the same timeframe. What could be the reason of the difference?

Best regards!

What timeframe are you testing? Also what are the details of the system that you’re testing? Some rules are more robust than others. If your rules are too tight, the system may be very sensitive to small changes in data. Data feeds from two different brokers aren’t 100% the same, even if it’s the same pair and timeframe, as there’s no central exchange in the forex market.

Hello,

thank you for the answer! The timeframe is M15, currency pair is EURUSD.

I have checked that time shift is not a problem. I mean, on the data from ForexTester it works well with +1 hour timeshift and without timeshift. On data from Alpari it works badly for both -1 hour timeshift and without timeshift. It is not a problem of symbol parameters as well.

What can I do in such a situation? Should I find parameters for which growth is stable for both historical data (ForexTester and Alpari)? Or how can I check which historical data is closer to Noble Markets data? How can I make it more likely that my system will work in similar way in real account, as compared with backtesting? Could tight stop loss be the reason of huge difference in resuls (I guess it could be)?

Regards!

Hello Johnyjj2,

what is your experience with this issue after one year left?

Regards

Hello!

Did You check, that there´s a shift of timezone between 2011-12 of the Alpari-datas (+1h)?
Check if this is a solution for Your problem.