Backtesting issue - Discrepancies in results

New comment
 

Hi,

I was backtesting mi EA on MT4, results were pretty good.

Yesterday evening I dowloaded data from the history center...now results (with SAME SETTINGS and on the SAME PERIODS) of backtesting are completely different!!!


Four main questions:

1 - How can this be possible?

2 - someone of you had the same experience?

3 - Which are the reliable results?

4 - how can be avoided for the future?


Thanks in advance for your contribution.

Thanks

Alex

 

If you download from the History Center, you get MQ demo data, not broker data

Always try & backtest on live data gathered (over time) and backup your history folder weekly as MT can sometimes trash the contents :(

Keep in mid that even a good backtest on live data is no guarantee of the same results in forward trading

This is particularly true of EA's that enter mid-bar, rather than on close of previous bar

FWIW

-BB-

 
Financialabs wrote >>

Hi,

I was backtesting mi EA on MT4, results were pretty good.

Yesterday evening I dowloaded data from the history center...now results (with SAME SETTINGS and on the SAME PERIODS) of backtesting are completely different!!!

Four main questions:

1 - How can this be possible?

2 - someone of you had the same experience?

3 - Which are the reliable results?

4 - how can be avoided for the future?

Thanks in advance for your contribution.

Thanks

Alex

Hi F,

when using the tester you should now that to use it only during week hours to test with diff spreads

all your diff results are because of the current spread that yr testin on. when you hit the start button on the tester, it will use the current spread of the market data

so when you are testin in weekend the market spread is really high, and it looks like yr bot is going bankrupt all the time in weekend

while you where sure the settings u used where working before while testing during workng hours

first of all dont just download history data, let yr metatrader collect it for you, and best is to use the real account

so when testin hit the start button at the spread you wanna test on

hope this helps you out

Greets JB

 
Financialabs:

Hi,

I was backtesting mi EA on MT4, results were pretty good.

Yesterday evening I dowloaded data from the history center...now results (with SAME SETTINGS and on the SAME PERIODS) of backtesting are completely different!!!


Four main questions:

1 - How can this be possible?

2 - someone of you had the same experience?

3 - Which are the reliable results?

4 - how can be avoided for the future?


Thanks in advance for your contribution.

Thanks

Alex



Its probably not your fault. All history data comes from Alpari (via MetaQuotes), and on Friday the data got totally messed up.

It worked great on friday but in the evening it was total chaos, and this has not changed to this time, monday night.

The problem seems to be that history data is badly corrupted more or less all over. It is however especially clear on EURGBP,

just look at this chart!

<http://home.online.no/~eikarlse/temp/Image2212121.jpg>

 

Guys,

thanks for your help.

To be sure to understand correctly the better procedure:

1 - keep metatrader open on all CURRENCIES and TIMEFRAMES that i want to use for testing

2 - backup history folder during week time (to avoid corruption)

3 - use backtesting on "live" collected data to obtain a more accurate results

4 - avoid to download data from the history center (correct?)

Thanks for your help

Alex

New comment