1 (edited by torinfx 2017-01-08 01:47:15)

Topic: How to solve: EA goes good in 12 years data but goes bad next 3 years

Hello everyone,
I used EaStudio to create strategy on time frame H4, from period 2001-01-02 to 2013-12-30 20:00.

It seems to me (but if I'm wrong please tell me :-)) that it's a good EA by looking at SQN, Sharpe Ration, Return/Drawdown, %Drawdown etc..

Optimization was initially made with no OOS, after a first backtest I changed to 50% OOS, but future results didn't change.

In the image attached you find the report.

Montecarlo Test (the one with 50% OOS) gave what follows:

Confidence      Profit    Profit / day    Max DD %      Return / DD    Trades    SQN           Win / loss
Original              17191    3.64                    8.37                8.39             118            4.53             0.75
10  %              17710    3.75                    8.14                8.64             116            4.72             0.76
95 %                      15134    3.20                    9.31                7.39             116            3.97             0.73
100 %              14913    3.15                    9.61                5.73             110            3.92             0.73

Then, even if maybe mr McKay wouldn't agree :-), I went on MT4 for backtest until dec2013 and to see how it behave in the next 3 years after the period test.

Backtest results was inside EAstudio Montecarlo range, but next 3 years was a completely loss: he lost 4.500$, 18/27 operations in loss, only 9 won.

Please, what are the conditions an EA must have in order to perform well in the period after the test too?

Thank you very much!
Angelo

Post's attachments

Schermata 2017-01-08 alle 01.26.07.png 88.82 kb, file has never been downloaded. 

You don't have the permssions to download the attachments of this post.

Re: How to solve: EA goes good in 12 years data but goes bad next 3 years

The numbers can fool us a bit so we have to be a detective

My son once wrote an ea that made millions until it got to 1990 then it went bad. and he could not find out why.

My first suggestion is to keep this one and create one or two more that actually trade in that period. They may not trade in the period that the original trades.

You seem to have a good quality strategy and it may well trade from today going forward. We never know what the market will bring us tomorrow.

Personally, I use a lot of strategies, try to get by without many losses, so I have long periods of no trades. hence mopre strategies to cover the dry spells...

Certainly you do not want to discard the good one.

If you change it. you will have more losses somewhere else....... to me, best to make additional.

Hope that helps.

BTW  I have not used the studio until today and my approach is to run it for 6 or 8 hours so as to get many strategies to work with.

Overnight I got about 300 that passed the MultiMarkets, I will see if I can get trades throughout the period with some of those.......

The new WorkFlow allowes us to make a gazillion strategies in a few hours, that will change the game tremendously.

An alternative to creating more in the same currency is to create for different currencies to see iof the gap can be filled properly.

Another alternative would be to use a lower time frame to fill in the gap.

With this work flow tool, there are tremendous possibilities.

My 'secret' goal is to push EA Studio until I can net 3000 pips per day....