Optimize vs Back Test

New comment
 

I've optimized an EA and gotten Input variable settings that are positive. When I use these settings to run a back test, the results are negative.

I've been thinking that the Optimizer is using 'minute' data while the back tester is using 'tick' data.

Is this correct?

Bill

 
oneoleguy wrote >>

I've optimized an EA and gotten Input variable settings that are positive. When I use these settings to run a back test, the results are negative.

I've been thinking that the Optimizer is using 'minute' data while the back tester is using 'tick' data.

Is this correct?

Bill

I don't think so, should be the same unless you've changed the settings.

I've noticed that some brokers don't use the same amount of data in optimizing and back testing, take a look at the results and see if the start dates are the same.

The negative results are likely due to the back test being run on out of sample data. always an issue with optimizing.

HTH

Keith

 
kminler wrote >>

I don't think so, should be the same unless you've changed the settings.

I've noticed that some brokers don't use the same amount of data in optimizing and back testing, take a look at the results and see if the start dates are the same.

The negative results are likely due to the back test being run on out of sample data. always an issue with optimizing.

HTH

Keith

I reviewed the History file (Tools/History Center) for the EURUSD and the minute data goes back to March 3rd, 2009. I changed the Start/End dates of my Optimization. Same results... Optimie = $330 Basck Test = -$376.

Any other suggestions?

Bill

 

> When I use these settings to run a back test, the results are negative

Could be there are trades open at the end of the back test which are closed by the tester regardless of profit/loss

Not sure that optimization does this?

-BB-

 
There's thin border between the optimisation and backtest. IMHO tester mostly was created for testing, optimisation - is option, with which I don't recomend you to play. Forvard test shows you possibilities of your EA. Use it.
 
sayfuji wrote >>
There's thin border between the optimisation and backtest. IMHO tester mostly was created for testing, optimisation - is option, with which I don't recomend you to play. Forvard test shows you possibilities of your EA. Use it.

Forward testing is running the EA in a Demo account and viewing the results, right?

It's funny (actually not very) that I run an optimization and use the results in regular Demo trading and the Eas JUST DON'T TRADE.

I don't understand that!

Bill

 

1OG

> I run an optimization and use the results in regular Demo trading and the Eas JUST DON'T TRADE

Any errors in th Journal?

EA & MT options & properties set to allow trades?

Where an EA enters mid-bar, there can be significant variation between backtest/optimization & forward trading

Additionally on live data, if the EA is sensitive to spread, the results will be very different, as backtest/optimization uses a fixed spread & obviously spread can vary on live (if not demo) data

FWIW

-BB-

New comment
Reason: