Can I use my EA live even if it has mismatched error and unmatched data errors. Thanks.

New comment
 
I am using M5. I don't need M1 data.
 

You can, though whether your backtest means much depends on several factors...

What was the modelling quality percentage?

-BB-

 
BarrowBoy wrote >>

You can, though whether your backtest means much depends on several factors...

What was the modelling quality percentage?

-BB-

Strategy Tester Report
MY_TEST_EA
FXDD-MT4 Live Server (Build 220)


Symbol EURJPY (Euro vs. Japanese Yen)
Period 5 Minutes (M5) 2008.03.25 06:50 - 2009.02.17 23:55 (2008.01.01 - 2009.02.18)
Model Every tick (the most precise method based on all available least timeframes)
Parameters MagicNumber=670312
Bars in test 66571 Ticks modelled 4583021 Modelling quality n/a
Mismatched charts errors 2999
Initial deposit 50000.00
Total net profit 109451.12 Gross profit 119305.00 Gross loss -9853.87
Profit factor 12.11 Expected payoff 1189.69
Absolute drawdown 326.75 Maximal drawdown 19961.73 (18.23%) Relative drawdown 27.42% (19807.74)
Total trades 92 Short positions (won %) 74 (83.78%) Long positions (won %) 18 (94.44%)
Profit trades (% of total) 79 (85.87%) Loss trades (% of total) 13 (14.13%)
Largest profit trade 4324.55 loss trade -1979.17
Average profit trade 1510.19 loss trade -757.99
Maximum consecutive wins (profit in money) 12 (16399.90) consecutive losses (loss in money) 2 (-2780.36)
Maximal consecutive profit (count of wins) 16399.90 (12) consecutive loss (count of losses) -2780.36 (2)
Average consecutive wins 7 consecutive losses 1

 

The number of errors relative to the number of trades and the n/a modelling quality means that confidence in the results is very low

With the low number of trades per year, will this run on a higher timeframe?

You might find you have better data quality there

This would help much more if your EA enters at the end of the bar rather than mid-bar...

If mid-bar you need seriously good data to have any relevance in your testing...

Hope it works out

Good Luck

-BB-

 
BarrowBoy wrote >>

The number of errors relative to the number of trades and the n/a modelling quality means that confidence in the results is very low

With the low number of trades per year, will this run on a higher timeframe?

You might find you have better data quality there

This would help much more if your EA enters at the end of the bar rather than mid-bar...

If mid-bar you need seriously good data to have any relevance in your testing...

Hope it works out

Good Luck

-BB-

By the time bar closes, it is too late. I need real time.

How can I test it without errors ? Do I need to download new data ?

thanks

New comment