EMA 25-50 Trading System (Detailed) | Page 110 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

10 Ema and 200 Ema cross system 6 replies

EMA crossover (ema 14 & 34) 4 replies

Please help code an EMA Crossover 12ema x 24 ema cross 1 reply

EMA's and MA's confiming small time correctoin ema's crossing 1 reply

EMA Close Crossing EMA High/Low Help Please! 2 replies

Attachments: EMA 25-50 Trading System (Detailed)
Exit Attachments

EMA 25-50 Trading System (Detailed)

  • #2,181
  • Aug 12, 2008 9:48am Aug 12, 2008 9:48am
Quoting tonirai
Did you read my post before yours?
For EURUSD, from 2008/01/15 until today:

Result: +10172 pips.

My result:
7301 pips (1-1-08 to 8-12-08)

For g/u
using your settings of 5/3/no/2.4
i get -1914 pips!!!

Something definitely wrong there
  • #2,182
  • Aug 12, 2008 9:57am Aug 12, 2008 9:57am
One question:
When you are trying past year(s), are you dragging your chart to visualize the beginning of the test period?

I ask this because, in my case, the 1H timeframe by default only has data since begin of this year... either you download the data, or you drag the chart to that date, to force MT to download the data.
Then you reload the Indicator (or switch between timeframes).

Corre, are you getting those results too in GBPUSD?
  • #2,183
  • Aug 12, 2008 10:18am Aug 12, 2008 10:18am
My results for 5/3/no/2.4 settings:

EURUSD:
7485 (2008/01/15 - 2008/08/12)
404 (2007/01/15 - 2007/08/12)

GBPUSD
419 (2008/01/15 - 2008/08/12)
5857 (2007/01/15 - 2007/08/12)


Results for no MACD and ATR 2.1

EURUSD:
7526 (2008/01/15 - 2008/08/12)
-153 (2007/01/15 - 2007/08/12)

GBPUSD
2136 (2008/01/15 - 2008/08/12)
3774 (2007/01/15 - 2007/08/12)

I've noticed that sometimes results are different when I change the pair in chart window without removing indicator. So all results where obtained on new charts window.
  • #2,184
  • Aug 12, 2008 10:34am Aug 12, 2008 10:34am
pavikogm,

Notice that optimal settings are only optimal for one particular pair...
Using it in different pairs won't show best results.
And be sure to have the past data in your chart (just loading it isn't enough, because it only loads data beginning at a particular (short) date.

Your first set of settings were my own for GBPUSD (strange result you had this year; "normal" last year);

I didn't use your second set of settings... for EURUSD my best was 5/2/False/2.3.
  • #2,185
  • Aug 12, 2008 10:40am Aug 12, 2008 10:40am
Yikes,
When I scroll back (all the way to 2001) then rerun it, I get -8k pips
interesting

this is on fxdd btw
  • #2,186
  • Aug 12, 2008 10:54am Aug 12, 2008 10:54am
Quoting tazmaan
Yikes,
When I scroll back (all the way to 2001) then rerun it, I get -8k pips
interesting

this is on fxdd btw
Meaning when you put the start date to 2001 until now?
I tried to scroll back the most, but for 1H chart, the maximum it goes is December 2005.

But for start 15/Jan/2006 until now, with the previous settings I used, it gives +18782 pips (average +- 4800 pips for 8 months).

But I suppose the more we scroll back, the less accurate will be the data, between providers.
  • #2,187
  • Aug 12, 2008 11:13am Aug 12, 2008 11:13am
No, I meant i scrolled back til 2001 so I had data back until then. I still only ran 1-1-08 til 8-12-08 and had that terrible number.
Not sure what's up
  • #2,188
  • Aug 12, 2008 11:40am Aug 12, 2008 11:40am
I've written EA with logic from Indicator V1.12. Unfortunately I've written it from scratch - it was faster than studying Zen Leow code. It's for testing purpose only till Zen is back.

Notes:
- ONLY FOR DEMO TESTING
- spreads are within TP and SL like it's with Indicator, so if we have break even it's counted as loss (we get a bit different statistic for win/loss, but I wanted to be as close to Indicator as possible)

I think there is another issue with Indicator:
Inserted Code
      // Getting the indicators values
      double ema =         iMA(NULL /*Current pair*/,
                              0 /*Current time frame*/, 
                              EMAPeriod /*period*/, 
                              0 /*No shift*/, 
                              1 /*EMA*/, 
                              0 /*Close*/,
                              i /*Shift from buffer*/);
We are getting EMA from Close price, so we are one hour ahead of market (in the future) I've noticed it with EA, because EA quite offen opened trade one hour later than Indicator - it's the cause.


The attachment below is EURUSD result from 2008.01.15 till 2008.08.12, settings 5/2/false/2.3, 0.1 lot size
Attached Image (click to enlarge)
Click to Enlarge Name: EMA25_EURUSD.gif Size: 22 KB
Attached File(s)
File Type: mq4 EMA_25_EA_test.mq4   25 KB | 389 downloads
  • #2,189
  • Aug 12, 2008 11:52am Aug 12, 2008 11:52am
Thanks Pavikogm!
So are you saying you fixed the issue of the one hour later in the EA you just made?

I will run this thru the strategy tester, and have it optimize the settings as best I can.
  • #2,190
  • Aug 12, 2008 11:56am Aug 12, 2008 11:56am
Quoting pavikogm
We are getting EMA from Close price, so we are one hour ahead of market (in the future) I've noticed it with EA, because EA quite offen opened trade one hour later than Indicator - it's the cause.

Thank you for that work

I'm going to be honest with you and admit I won't be testing it so soon:
I like to take things step by step... and right now I'm concentrated in finding the best parameters for several pairs.
I'll post an approach for this next.

But about the EMA issue:
It shouldn't be a problem, because beside all other checks at close of previous candle, the EMA condition is:
"if price opens above/bellow EMA close on current candle";
So the open price is one alone (doesn't change);
even though the EMA close is moving, if at the openning of the new candle the open price is above/bellow the EMA, it should instantly trigger a buy/sell...
So where's the problem?

By the way, with what leverage was this account running?
  • #2,191
  • Aug 12, 2008 11:57am Aug 12, 2008 11:57am
Yes. MetaTrader "Tester" tool for EA simulates real conditions (including spread, swap points) and we can't have situation of reading from the future.

For Indicator we need to change code to:
Inserted Code
// Getting the indicators values
      double ema =         iMA(NULL /*Current pair*/,
                              0 /*Current time frame*/, 
                              EMAPeriod /*period*/, 
                              0 /*No shift*/, 
                              1 /*EMA*/, 
                              0 /*Close*/,
                              i + 1 /*Shift from buffer*/);
  • #2,192
  • Aug 12, 2008 12:03pm Aug 12, 2008 12:03pm
Quoting pavikogm
Yes. MetaTrader "Tester" tool for EA simulates real conditions (including spread, swap points) and we can't have situation of reading from the future.

For Indicator we need to change code to:
Inserted Code
// Getting the indicators values
      double ema =         iMA(NULL /*Current pair*/,
                              0 /*Current time frame*/, 
                              EMAPeriod /*period*/, 
                              0 /*No shift*/, 
                              1 /*EMA*/, 
                              0 /*Close*/,
                              i + 1 /*Shift from buffer*/);
I think I get what you mean:
It's an issue for the Indicator, not for the EA, right? (I though you were saying it was for the EA).

We can put it like you say (Close; i+1), or "Open; i"... these two should be the almost the same, right?
  • #2,193
  • Aug 12, 2008 12:05pm Aug 12, 2008 12:05pm
Quote
It shouldn't be a problem, because beside all other checks at close of previous candle, the EMA condition is:
"if price opens above/bellow EMA close on current candle";
There is problem, look at this strange example for simplicity (4 candles from 0:00 to 3:00):
Open Bar Price: 1 1 1 1
Close Bar Price: 2 2 2 4

EMA is about (exact value is not important for example):
2 2 2 3

but value close for 3:00 candle will be know at 3.59. You are checking, if at 3:00 time, open price is higher/lower than EMA at 3:00 (which is calculated form 3:59 value).

Quote
By the way, with what leverage was this account running?
Money mangement was turned off. Lot size was constant - 0.1 lot. Initial deposit was $10000. Leverage can be anything. For 1:100 we require $100 maring.
  • #2,194
  • Aug 12, 2008 12:54pm Aug 12, 2008 12:54pm
I had a problem in GBPJPY too, some kind of software blocking:
I kept changing parameters, but neither the signs in the charts nor the counter changed;
I removed the Indicator and reloaded it, and then it was OK.


Guys,
I know now there's an EA, but for those who like to do some manual testing, here's what I think I'll do, as an approach to choosing best parameters for different pairs, starting always with MACD=False and the rest as defaults:

1) I play around Multiplier=2.0 (+0.1, -0.1, etc) to find the best one;
2) When I found the one or two best values, I play with Stoch_Slowing & Stoch_MA_Method, confirming which Multiplier is the best - Then I stick to this Multiplier, don't change it anymore;
3) After that, you make a matrix (for example) of 12 possibilities -> 3 Stoch_Slowing (5,6,7) * 4 Stoch_MA_Method (0,1,2,3), and change the date for each of the 12 combinations, noting the results;
4) You find the best parameters when you get higher sum of pips for all years (I use 2006, 2007 & 2008), or when you find similar number of pips in all years (I prefer the last one, as it is better in 3 years, to have “4k+6k+5k=15k” than “10k+1k+5k=16k”
  • #2,195
  • Aug 12, 2008 1:13pm Aug 12, 2008 1:13pm
Ok, my first run at optimization, this is for EUR/USD from 1/1/08 to 8/12/08

First run of optimization, only letting it optimize
the stoch slowing (4, 5 or 6)
the stoch ma method (0,1,2,3,4,5)
and the ATR (from 1.0 to 3.0 by increments of .1)

Also, did both MACd true and false (no change substantial change in results)

Best settings for Macd=true
stoch 5/1 atr 2.6 = +$5636 (so about 5636 pips including spread)

Best settings for Macd=false
stoch 6/3 atr 3.0 = $5876 (so about 5876 pips including spread)
  • #2,196
  • Aug 13, 2008 9:25am Aug 13, 2008 9:25am
So...
What means all this silence?
Is it that everybody is absorved in testing the EA or...
Is it that people found some kind of problem?

Well, I also tried the EA 2 or 3 times... but it seems that EAs or Strategy Testers aren't my thing!
I'm never able to make a test until its end (with this or the previous EA)... there's always an error that stops the test, like "not enough money in the account".
I think I simply don't understand them enough and don't know how to configure it and my account the best way.

But I've made manual optimizing and calculations tests for 6 pairs until now, considering the beginning of 2006 until now, using the technique I told yesterday.

After a first result in a pair (with certain settings) in the first 8 months of each 3 years (from the matrix I mentionned), I choose 1 to 4 best combinations of Stoch_Slowing & Stoch_MA_Method in the 3 years;
I work my way back starting at the worse of those years (generally 2006) for each combination (now counting for the all year), changing the Multiplier around it's inicial value and noting the results;
Then I choose the best combination (stability over time & higher pips).

Some findings:
2006 is almost always the worse year;
The end/rest of the year (after the intial 8 months of the first test) seems to performe worse;
Best results with EURUSD & GBPJPY;
Average results with GBPUSD, EURJPY & GBPCHF;
Worse (not good enough) result with USDJPY.

I know it's not as accurate testing as a computer one is supposed to be, but it fits my purpose.

Overall, I'm pleased with the results, and will test 2 or 3 more pairs.
(I work for the next 2 nights, and have a wedding next Saturday, so don't know when I will)
  • #2,197
  • Aug 13, 2008 11:27am Aug 13, 2008 11:27am
Quote
I know now there's an EA, but for those who like to do some manual testing, here's what I think I'll do, ...
Quote
... What means all this silence?
Is it that everybody is absorved in testing the EA or...
Is it that people found some kind of problem? ...
Tonirai, EA hasn't changed much. All your proposed steps for testing are perfectly valid also for EA and demands some manuall interactions. Only presentation changes - you get charts and graphs with EA.

I think everyone is waiting for your findings about best settings I haven't much free time to play more with Indicator and EA today

Regards
  • #2,198
  • Aug 13, 2008 11:58am Aug 13, 2008 11:58am
Quoting pavikogm
Tonirai, EA hasn't changed much. All your proposed steps for testing are perfectly valid also for EA and demands some manuall interactions. Only presentation changes - you get charts and graphs with EA.

I think everyone is waiting for your findings about best settings I haven't much free time to play more with Indicator and EA today

Regards
I don't believe I have such influence in this thread

Seriously now, I will post then later (I'm going out to work in minutes)...

But since it's obvious we all have different brokers, which bring different results, I'm sure "my settings" won't be the best ones for everybody
And don't forget it's a manual approach, with all limitations that means (I'm sure there must be better configs than the ones I found, but it's hard to find them by manual testing).

The last few hours, I've been reading tutorials about Strategy Tester, and how to maximize its performance.
Interesting thing I learned was to maximize the "bars in history" amount; also to download 1M tick history and to import that to all the other timeframes.
But what I meant to learn was to use the configs in it (inputs & optimization tabs)... I'm still short on that.

See you all soon,
Toni.
  • #2,199
  • Aug 13, 2008 12:00pm Aug 13, 2008 12:00pm
I'm ready to test the EA, it would be great if we can have an EA as configurable as the last indicator.

(I'm trying the best settings for every pair these hours)
  • #2,200
  • Aug 13, 2008 1:31pm Aug 13, 2008 1:31pm
Toni, I can give you some assistance with that if you require. Let me know via pm if you can't figure it out.
3 traders viewing now