Maximal drawdown in back testing - what exactly is it?

[nsd]

Folks,

 

I am confused.  Can you please explain for me Maximal and Absolute Drawdown.  I thought I understood but in my testing it shows a gross loss of 63 and a maximal drawdown of 4030 and absolute drawdown of 1715 but it never drops more than 21 below the opening balance ... how can that be?

 

See the attachment please.  (Gotta love the profit factor)   :-) 

 

 

 

[Clerin]

https://www.mql5.com/en/articles/1486

https://www.mql5.com/en/articles/1403

Notice the comments at the bottom of each of those pages. Since build 204, drawdown is now calculated with equity rather than balance.

https://www.mql5.com/en/forum/103174

EDIT: You probably already know this but, since it is related to the gross loss in your question, that profit factor is jumping off the page as ... somewhat higher than one might reasonably expect ... of a live trading system. A profit factor that high would tend to point towards over-optimization. It is possible to check out the profit factors of the winning participants in the last automated trading championships to get an idea of profit factors from live trading results which many people consider to be healthy.

[nsd]

clerin6:

https://www.mql5.com/en/articles/1486

https://www.mql5.com/en/articles/1403

Notice the comments at the bottom of each of those pages. Since build 204, drawdown is now calculated with equity rather than balance.

https://www.mql5.com/en/forum/103174

EDIT: You probably already know this but, since it is related to the gross loss in your question, that profit factor is jumping off the page as ... somewhat higher than one might reasonably expect ... of a live trading system. A profit factor that high would tend to point towards over-optimization. It is possible to check out the profit factors of the winning participants in the last automated trading championships to get an idea of profit factors from live trading results which many people consider to be healthy.

Thanks, I had read the links you sent but clearly not read all the way down to the bottom.  It makes sense on equity.  I'll have to re-visit my other test results now I understand better.

 

As for over optimization, perhaps, but actually I was simply forward testing an EA I had optimized to be profitable each year since 2008.  The forward test was for 2012 and the profit factor made me laugh, but it is not designed for 2012.  It is designed for all years since 2008.  Since 2008 it has a 83% win ratio and profit factor of 2.91 - so agreed 2012 is just a weird outlier.  But if that translates into a demo account and live account we will see..........   Monte Carlo or bust as they say.

 

Thanks again.