Thread: Idea for 121 EA

Hi Jonty172,

I'll have a look back on my platform, hopefully my M5 data goes back that far. Could you just clarify your comments around the EURJPY. Did you say it won the trade then lost it? Just in case you aren't aware, you need to be careful of the strategy testers spreads when backtesting over the weekend. Sometimes on the last tick of the week the broker sets them sky high and they stay there when backtesting over the weekend. It can drastically affect results. You can get scripts that allow you to set the backtester spread. There's plenty of articles on it.

Cheers
Paul

Originally Posted by Jonty172

@ PAUL It is on Gbpusd 5min 2010 Jan 3rd the 1st trade enters jan 5th and is a small looser of around 17-25 pips then no more trades were taken for the rest of the period to March 2012. It was really strange too, I had the exact same thing with EURJPY and it won around the same entry time (or very close to it) + 37 ticks, then I tried again because it never took a further trade, and it lost the same trade

Hope all this makes sense and will help you modify this great looking EA and help us all out at the same time

@ PAUL Thats correct it won then lost it! I dont think that was the problem of the spread changing over the weekend as I have downloaded the charts for the last 10 years and running them offline anyway at 90%. Now its even weirder, the cable trade it always takes on the 4th/5th jan 2010 isnt happening. Let me ask, all those settings u recommended in an earlier thread, are those what u r running or do I need to modify them at this stage?

@Jonty172:

I don't understand how it can win a trade then lose it, surely once the trade is closed it's closed? Or do you mean you ran the EA once and it won, then the 2nd run it lost? Regardless of your datasource the backtester always gets it's spread in realtime from the broker, so at the weekend when the market is closed it will just use the spread of the last tick on friday. Historical data doesn't include spread (or effectively ask prices). This is one of the major floors with MT4. there's heaps of forum threads on it, this is just one Historical data in MT4 includes spread? - MQL4 forum When you back test at the weekend every tick uses the same spread value. backtesting during the week is better as it will get the current spread of the broker. It's still not ideal as you're still effectively applying a single spread across every bar in your backtest. The tester spread value can be found by clicking the 'symbol properties' button in the strategy testing window. It may not be wildly high, but it will definitely be different to when you ran it during market hours.


Anyway, my broker's data doesn't go back that far I can only get back to 18/04/2011. I ran the EA from there and had 2 trades. one winner and a loser as per below. The loser was a pattern that just kept going, so no save by the exitzone. Once we start building confirmation techniques in to the EA I'm hoping we can avoid these. So below are 2 trades the EA took, one after the other, so the EA seems to be working after the first trade. Maybe if you can switch on the screenshot functionality and post a screenshot of the trade (please turn on ask line too) we can try and understand what's happening.

Cheers
Paul

This one is also a nice example of a non-failing pattern moving to profit well before the failed pattern exitzone



This was the loser

Forgot to ask, did you see patterns appear but it didn't enter a trade, or did you just see no more patterns after the first one?

Important to know, as one is the EA (not trading when it should be) and the other could be an issue with the indicator


EDIT: I just use the default. Just make sure you have enableabcd and enable50 set to true if you want to trade both patterns.

The EA closed this at breakeven, but had the entry been tighter it would never have trigger the stoploss move and eventually went on to hit the SL. As you can see closing at the exitzone (purple arrow) it would have actually closed in profit

Hi Paul, u r totally helpful! Ok so I see the patterns but it wont trade them all, I just did the same time you showed on the picture and it gave me the same trade. Looks like progress. In terms of the Eur JPY that was a re-test where it gave me the same results.
I will let you know more info or if I have any questions soon but it looks good, the test on cable is still running but it has not done a further trade after the copy of yours in May 2011 and I am now at July 5th.

Only thing I can think of if your using the filters MinProfit_Points or MinRewardRatio. MinProfit_Points default to 100 (10pips), if the pattern profit line is less than 10pips then it won't enter. likewise if sl/tp doesn't meet risk=1/reward=MinRewardratio then it won't enter.

Originally Posted by Jonty172

Hi Paul, u r totally helpful! Ok so I see the patterns but it wont trade them all, I just did the same time you showed on the picture and it gave me the same trade. Looks like progress. In terms of the Eur JPY that was a re-test where it gave me the same results.
I will let you know more info or if I have any questions soon but it looks good, the test on cable is still running but it has not done a further trade after the copy of yours in May 2011 and I am now at July 5th.

Yeah min profit is 100 and min reward is at 0, shall I change them???