To "Mirror" the Edge

[charvo]

Hi, I posted this question before in "bug" forum. I think I'd better post it here again with better elaboration of the issue that I have now.

This seems an interesting topic.

As we know, if there's a consistent loser system, I can simply reverse the long with short, and vice versa, to get a consistent winner system, under the assumption that the "consistent loser" is NOT due to swap&spread costs. In other words, the "loser" is statistically significant.

now i'm pretty sure my system (both .dpr and .dll attached) is STATISTICALLY consistent losing ( much more than merely losing spread&swap) , so I switch all buys to sells, and switch all sells to buys. I also use takeprofit vs stoploss = 1:1 to make sure the reversed system SHOULD take profit when the original system stoploss.

As a result, I should've had a consistent winning system, shouldn't I?

but the FT2 cannot exactly replicate the opposite entries in my experiment. The only cause that I can think of is the swap&spread, but it really does NOT look like so.

now have a quick idea of the system:

original idea: http://www.forexfactory.com/showthread.php?t=262936

basic operation: the system keep watching 10 pairs, and by a time interval (e.g. 1 hour), the system will enter a position based on the leading pair (the pair rise/drop most). the system keeps adding positions following the leading pair until the overall profit (or loss) target is reached. Then all positions will be closed, and the system wait for next day to start it again.

so if you care to drag my system into your FT2 to run (with default setting), you will see every hour (or whatever TF you select) the system enters a 0.04 lot position of the leading pair, then maybe after 10+ hours, these 10+ positions may reach the target of $200(+/-), then the system will close all of them.

You may run it for a couple of months data, and it usually will give you a loser performance.

then you can switch the "Reverse Gear" option to true, and re-run it for the same period. Since the takeprofit vs stoploss = 1, this reversed system should exactly "MIRROR" the original system's performance as well as trades!

But it didn't! and this is my question now. Why can't I "mirror" my system trades by switching the long and shorts?

I am pretty sure that I see the "negative edge" from my orignal system's performance. and this "negative edge" itself is as golden as positive edge in system development. All left to make a profitable strategy is just to mirror its trades, but .......

I hope that anyone who feels interested may give a try on the attached system and share your insights. If you do try, try to generate your ticks data with swap & spread as 0 so that the reversed system and the original system (is supposed to) share same entry prices.

My next step is to re-generate all my data to be swap&spread-free (make them all zero). if that does not work, then I might guess it is FT2's issue/bug......

[FT Support]

Hello,

The task which you are trying to do is not as simple as reversing Enter conditions only. It is also about Exit conditions.

Please try to remove ALL "Exit" conditions from the strategy and leave only Enter with SL and TP after that reversed results will be more consistent with non-reversed results. If this does not work then post your test code (without exit conditions) here and we will take a look.

[charvo]

thanks, I've solved the problem. i set all spread and swap as 0, now the trades may accurately be reversed.

however, i do find a bug/problem of FT2.9. after it runs a while, the test may generate different trades under same settings (i like to run same test repeatedly). if i restart the FT2.9, the different trades will be gone. the trades will be same again. is this a memory problem?

it has happened many times to me, not sure if this is an issue that's been reported.

Hello,

Please try to remove ALL "Exit" conditions from the strategy and leave only Enter with SL and TP after that reversed results will be more consistent with non-reversed results. If this does not work then post your test code (without exit conditions) here and we will take a look.

[Phil_Trade]

however, i do find a bug/problem of FT2.9. after it runs a while, the test may generate different trades under same settings (i like to run same test repeatedly). if i restart the FT2.9, the different trades will be gone. the trades will be same again. is this a memory problem?

it has happened many times to me, not sure if this is an issue that's been reported.

Hi

Be carefull to reset all variables, array etc in ResetStrategy otherwise you could have different result between each run

[charvo]

Thanks Phil, I didn't realise this.

Hi

Be carefull to reset all variables, array etc in ResetStrategy otherwise you could have different result between each run