Back testing EA: different result on different broker account

 

Hello Everyone,

I have done backtest for my EA on H4 using Strategy Tester, Control Point setting. I have few questions and hope you can help.

1. Using broker historical data, the results on Fxpor, Peperstone, FXDD, Axitrader are not the same. The shape of the balance Graph is similar across these broker. However the returns are not with the range from 95% to 70% .

I wonder if it is normal phenomenon we face when doing the back testing using broker historical data.

2. What is the cause for such difference?

3. Since my EA open order right after the close of H4 candle, how reliable is the result when I use the Control Point setting?

Thank you for your help.

SCFX

 
Because your brokers are probably in different timezones and thus have different server times, so the H4 bars will not look the same. Read this thread posted today: https://www.mql5.com/en/forum/147293
 
scfx:

Hello Everyone,

I have done backtest for my EA on H4 using Strategy Tester, Control Point setting. I have few questions and hope you can help.

1. Using broker historical data, the results on Fxpor, Peperstone, FXDD, Axitrader are not the same. The shape of the balance Graph is similar across these broker. However the returns are not with the range from 95% to 70% .

I wonder if it is normal phenomenon we face when doing the back testing using broker historical data.

2. What is the cause for such difference?

There is no central market for Forex so each Broker's prices can be a little different . . . what was the spread for each run you did ?
 

Good morning,

On the Tester, I choose Spread: 2 for all the test.

SCFX

 
scfx:

Good morning,

On the Tester, I choose Spread: 2 for all the test.

SCFX

Good evening . . . it's probably just a matter of the Brokers having different history data.
 

different brokers use different price feeds, one time several of us on another forum took screenshots of our charts at the end of trading on a Friday, comparing those screenshots revealed the highs and lows during the hours leading up to close of trading varied by as much as 15 pips between brokers.

 

Thank you for your comments.

I wonder if when it put spread =2 for GU testing, is that reasonable spread to expect on a normal day?

I know certain broker like FXpro charge really high spread on even GU (say 3 pip).

Thank you,

SCFX

 
scfx:

Thank you for your comments.

I wonder if when it put spread =2 for GU testing, is that reasonable spread to expect on a normal day?

I know certain broker like FXpro charge really high spread on even GU (say 3 pip).

Thank you,

SCFX


In the tester, the spread is in points, so for a normal 5 digit broker, you would enter 20, not 2