My current EA | Page 2 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Looking for indicator: alert when current price equal high/low of current bar 26 replies

US Current Account / US TIC 43 replies

Current Candle Closed flag 3 replies

Current Daily Open 0 replies

Pure Price Action: Predicting the Current Move 9 replies

Attachments: My current EA
Exit Attachments

My current EA

  • #21
  • Jan 12, 2012 11:16am Jan 12, 2012 11:16am
First I'll try to set it up on a demo, just to watch if my broker accepts everything as it should .. currently I cannot Send S/L and T/P right with the order, but need to send an extra modify - that's annoying and i have to implement that first.

Also I need to know how good the orders will be handled when sending directly instead of Strategy Tester.

If evverything works as expected, and the EA works well on demo, I will surely setup an Live Account and blow 500 bucks in just to see if it works.

No problem if I lose it, is a cheap price for wisdom :-p
  • #22
  • Jan 12, 2012 11:48am Jan 12, 2012 11:48am
Quoting mr2
I will try that out surely.. but maybe I just raise a demo account and let it run and see what happens :-)

Is it right, that MT4 does not use the real tick-data? So it is possible that some peaks are hidden during testing? I have no further information about this issue.

But I feel confident, that it's not really important because of the way my EA works .. it is not really dependent on perfect tick data .. there are only pending orders involved.

Sure there can be many things happen and there is still a hand of luck involved..

As it looks,...
Real time price feed (demo and real) is in ticks in all brokers, but most of them do not store them in tick.. only in minutes. Therefore, when you download 1m data from metaquote directly from your mt4 you get 1m data with open, high, low and close price. When you run that in mt4 tester it does not mimic the live trading.

Extreme Example to reflect the fault ... price start in zero, goes to 50, then goes to -80 , then goes to 49 then goes to -79 and close at 30. ..all in 60 seconds.

The tester only go once to the high an low in a fashion based in the close price. in this case 30. Not according to the real price action. Also does not go back a second time to certain price that is already registered .. in this case will not go back to 49 and -79 because is already registered when went thru 50 and -80. So in some cases if you have pending orders you could show 2 winners but in reality they were losses because the the tester inverted the timing process of the real price action.

Now. 1m built with tick and use in mt4 recreate everything the market did in the timely fashion that happens.

Now , another issue. I see you are not even getting 90% quality tests, which you should get in high time frames.

Maybe is because you simple downloaded 4h data from metaquote and not 1m and then use that 1m to convert to 4h. That would produce the same fault as explain above for the tick vs 1m

I hope that helps


.
  • #23
  • Edited 12:30pm Jan 12, 2012 12:08pm | Edited 12:30pm
Thank you very much for the detailled explanation.

I was already aware of the mentioned faults, but they should not matter - it's still a bunch of numbers.

I was simulating by converting the M1 (when I am right .. it says "collecting m1, m5 etc") .. but because I am not a pro with using that, i've downloaded every data available at metaquote - I dont think they send me 1M Data for the last 20 years - which would be a very large download.

If I backtest only some days I get a much better modelling quality - I think because this data is truely available locally.

The EA is currently running on demo account (I am watching it right now) - and set off his first orders - I've just used a account I already got with a balance of about 150 $ :-) So let's see what happens .. need to wait now .. H4 is not the fastest Timeframe xD

The Order Sending (because I need to send them and right after do a modify with the stops) is not as fast as I believed in the first place .. maybe that makes it really unsuitable for more lower timeframes ..

If this one proofs its function I will try to model my concept on lower timeframes but I am not very confident, because movement is too low and to noisy which results in much more risk .. currently nothing under H4 is working for my EA because OrderCost/Movement Ratio is too low

Another Question:

I have opened the chart, setted it to H4 and then attached the EA to it .. is there a safer way to attach it to a timescale? I expect, if I change the Timescale now on the chart, this will affect the EA as well. True? So can I lock the EA to a specific timeframe outside of code?
  • #24
  • Jan 12, 2012 1:07pm Jan 12, 2012 1:07pm
Quoting mr2
Another Question:

I have opened the chart, setted it to H4 and then attached the EA to it .. is there a safer way to attach it to a timescale? I expect, if I change the Timescale now on the chart, this will affect the EA as well. True? So can I lock the EA to a specific timeframe outside of code?

In my humble opinion I don't know any other way to attach an EA to a timescale, except a tricky way, that would be having two instances of the MT4 terminal, one with only one chart with the EA attached and if you right click on the timescale tab (on top M1 M5 M15...) you can customize and you can only let the timescale you wish...
  • #25
  • Jan 12, 2012 1:31pm Jan 12, 2012 1:31pm
well, that's not so bad .. as long as I do not transfer the EA to a server I will let it run on a netbook, so I can always see how it's doing.

thx anyway
  • #26
  • Jan 12, 2012 4:48pm Jan 12, 2012 4:48pm
Quoting mr2
-
Another Question:

I have opened the chart, setted it to H4 and then attached the EA to it .. is there a safer way to attach it to a timescale? I expect, if I change the Timescale now on the chart, this will affect the EA as well. True? So can I lock the EA to a specific timeframe outside of code?
Yes, changing the TimeFrame after you attach the EA will affect the EA.

But if you want to monitor same symbol in another TimeFrame , just open another chart. You can change that aditional chart to any TimeFrame you want and will not affect your EA.

If you want safety, so your EA will execute trades according to only 1 TimeFrame , you can hard code that. Then does not matter to what TF you attach or change.

Was that your question?.
.
  • #27
  • Edited 4:41am Jan 13, 2012 3:29am | Edited 4:41am
Yes, thank you :-) So, the EA is attached to 1 Chart-Window or a pair? I hope it is attached to a window .. because if I open another chart of the same pair it starts with wrong timescale ..

The EA is now running since yesterday night. on H4 and played 4 candles as it was supposed. It's more or less breakeven on profit and loss, but that's normal in this short period.
Update: EA reached profit area @10:30 GMT Works as supposed, more losses than wins, but profitable

I will backtest this day later in Strategy Tester, just to know if the results are equal to live tick data.
  • #28
  • Jan 13, 2012 9:12am Jan 13, 2012 9:12am
Quoting mr2
Yes, thank you :-) So, the EA is attached to 1 Chart-Window or a pair? I hope it is attached to a window .. because if I open another chart of the same pair it starts with wrong timescale ..

The EA is now running since yesterday night. on H4 and played 4 candles as it was supposed. It's more or less breakeven on profit and loss, but that's normal in this short period.
Update: EA reached profit area @10:30 GMT Works as supposed, more losses than wins, but profitable

I will backtest this day later in Strategy Tester, just to know if the results...
You can attach as many EAs as you want to as many windows your RAM allows you, each chart is independent BUT they must have different Magic Numbers, if they have the same MN they will get confused and manage all the trades with the same MN.
  • #29
  • Edited 12:06pm Jan 13, 2012 10:39am | Edited 12:06pm
That's usefull thank you :-)

Because I am not a patient person and my current EA is a slow one .. I played around even more. For using smaller Timeframes I needed a better logic for enty point.

My EAs uses partly some sort of previous high/lows currently - that was working pretty good for it's purpose .. so I tried to improve that .. get better signals to filter out small bounces..

I've made up such an EA and ended up currently with a ~80-90% wins (backtested on m1 but needs very good margin sadly .. ). I started it in DEMO some minutes ago .. sadly .. it generates only ~20 signals per week .. I am curious if it gets any signal today. For the first this second EA is standalone - if it works in realtime, I'll try to merge the two
  • #30
  • Jan 13, 2012 1:13pm Jan 13, 2012 1:13pm
Little update about the last 24 hours the EA was running: It was shortly positiv and ended with loss cause of a bounceback :-) Maybe I can tweak it a little, but to be honest, that was also in backtesting and doesn't matter in the long run.

I backtested the EA with the same time as it was live trading in Strategy Tester (using M1 Data) and the results was the same as during the live session.
That's still not very accurate but next week, the EA will run again ;-)
  • #31
  • Jan 17, 2012 3:38am Jan 17, 2012 3:38am
Quoting mr2
Is it right, that MT4 does not use the real tick-data? So it is possible that some peaks are hidden during testing? I have no further information about this issue.
Yes, this is correct, MT does not have tick data, floating spread, etc. Moreover, there are a plenty of holes in standard MT4 data so i'd not recommend it.
I have a program which allows to import good broker-specific tick (with floating spread) and minute data into your MT4 terminals, if you need this program - let me know.
  • #32
  • Jan 17, 2012 6:04am Jan 17, 2012 6:04am
At most I don't like the fact, that I cannot see the tick data in the chart .. i can only see an abstraction of it by candles .. that's good for me and how human brain works - but that's a bad base to write an EA on it.

I will try to write a small standalone program by my own to see and analyse this tick data .. the easist would be to just write an EA which exports real real-time tick data during one day and use this to analyse different strategies.

Second type of approach would be to write an "lagging EA" with no S/L at all completly handled by the EA only on closing candles. Not very precise but excellent in backtesting ;-)
  • #33
  • Jan 17, 2012 8:00am Jan 17, 2012 8:00am
[quote=mr2;5301591]At most I don't like the fact, that I cannot see the tick data in the chart .. i can only see an abstraction of it by candles .. that's good for me and how human brain works - but that's a bad base to write an EA on it.



You can see ticks in mt4..
Go to: view/market watch/. highlight the symbol you want to see and click "tick Chart"
  • #34
  • Jan 17, 2012 8:02am Jan 17, 2012 8:02am
I know, but I cannot scroll them, nor zoom them and it shows only the last 20 or so.

Don't know if I can change some setting, so it shows me the current day, but I don't think so.
  • #35
  • Jan 17, 2012 9:20am Jan 17, 2012 9:20am
Quoting mr2
I know, but I cannot scroll them, nor zoom them and it shows only the last 20 or so.

Don't know if I can change some setting, so it shows me the current day, but I don't think so.
I think that is all what you can see.
I never look at the tick charts.

.
  • #36
  • Last Post: Jan 17, 2012 9:56am Jan 17, 2012 9:56am
I've found this little indicator, maybe is what you are looking for
Attached File(s)
File Type: mq4 Tick_on_Chart.mq4   2 KB | 344 downloads
0 traders viewing now