EAs under construction II | Page 7 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

trading system under construction 4 replies

TradingRightByListen ingAndSpeaking (Trblas) Under Construction! 5 replies

Eas under construction III 24 replies

EAs under construction 68 replies

Am i under under-capitalized? 6 replies

Attachments: EAs under construction II
Exit Attachments

EAs under construction II

  • #121
  • Jul 8, 2010 10:31am Jul 8, 2010 10:31am
Quoting Haishang
Jiva,

I tried to have same parameter as you do (see attached excel spreadsheet with all details), again not really succesful. Before I start testing my own parameters I try to gain common ground by making sure we are talking exactly about the same thing. That is the reason why i want to make sure that if we use same parameters we receive exactly the same results. Please note that I am on alpari uk live, that means my offset is GMT-2 in order to get same start time. is that correct and corresponds to your offset? I think so but for safety reasons...
Alpari UK is +2
  • #122
  • Jul 8, 2010 10:40am Jul 8, 2010 10:40am
Quoting Haishang
Jiva,

I tried to have same parameter as you do (see attached excel spreadsheet with all details), again not really succesful. Before I start testing my own parameters I try to gain common ground by making sure we are talking exactly about the same thing. That is the reason why i want to make sure that if we use same parameters we receive exactly the same results. Please note that I am on alpari uk live, that means my offset is GMT-2 in order to get same start time. is that correct and corresponds to your offset? I think so but for safety reasons...
saving the report in mt4 only makes the graph easily accessible, unfortunately the details of the settings do not save in an easily accessible file for downloading. What I do is use the edit function to highlight the section of the report that gives the details of the settings, copy it and paste it directly into the post. Seems to be getting the job done.
I want to see your settings for these results. Could you do that? Post the settings using the edit copy and paste function?
Thank you.
(Some of us may not have access to excel spread sheet format)...
jiva
  • #123
  • Jul 8, 2010 10:41am Jul 8, 2010 10:41am
Quoting sanyhuis
What I don't understand.....how do you have a TSL of 300 on a trade with a TP pf 50 ? Am I missing something? I know it's Friday. Maybe that's it.
Thank you.
sandra
Sandra,

as far as I understand ( and this can only be rudimentary as you see from my backtests that I am far from Jiva) there is the following:

EUR/CHF
T/P 1 15 pips
trailing S/L 1 140 pips

.....means that if 1st order is filled with a t/p your second order basically is free to run as it will only hit the trailer if there is a huge correction, i.e. intervention

T/P 2 1000 pips
trailing S/L 200
That is too far away to be ever reached and hence it will expire

S/L 140 pips
That is s/l for first position in case it goes against you

Does this make sens in case of EUR/CHF for you?
  • #124
  • Jul 8, 2010 10:44am Jul 8, 2010 10:44am
Quoting sanyhuis
I don't understand how you can have a TP = 20 and a TSL of 140
& TP = 50 and a TSL = 300. And a SL = 140.

I can't put that together. Is it me and because it's Friday? Or is it not possible?

sandra
By using a trailing stop that is larger than the target it invalidates the trailing stop. That is the idea for these settings. No trailing stops are used. They are invalidated by the larger than target settings.

Once the 1st profit is reached the stop for the 2nd position moves to breakeven and fixed there. The idea is to let the 2nd position catch a big move if not it will stop out at break even. This is one of the parameter under testing. If it is more advantageous to give a set profit target for the 2nd position, say 50 pips that it can reach fairly often, or to try to let it catch a big move of perhaps 200 or 300 pips every now and then. That may be different for different pairs also.
  • #125
  • Jul 8, 2010 10:51am Jul 8, 2010 10:51am
Hi Jiva,

please find below the settings for my 2 preliminary testruns..... and attached as well the excel sheet which makes comparison between the various settings easy to differentiate! Thank you so much for your patience...I am a slow runner....but hopefully with enough endurance

Roman


EUR/CHF

EURCHF (Euro vs Swiss Franc) Period4 Hours (H4) 2009.07.01 00:00 - 2010.06.30 20:00 (2009.07.01 - 2010.07.01) ModelEvery tick (the most precise method based on all available least timeframes) ParametersiMagicNumber=532753; GMTOffset=2; TradeStartHour=1; TradeEndHour=24; Lots=1; TakeProfit1=15; TrailingStop1=140; TakeProfit2=1000; TrailingStop2=200; StopLoss=140; PendingPips=25; ExpireCandles=1; DelayOrder2=0; CheckSlope=false; DSS_name="Robby DSS Bressert Colored with alert v1.1"; DSS_EMA=8; DSS_stoch=13; Use_T3MA=true; T3MA_name="T3MA"; T3MA_Periods=400; T3MA_VolumeFactor=0.7; Channel_EMAPeriod=3; Channel_EMAShift=0; ShowComment=true; EXP_MA_1="-------- MA settings --------"; EXP_MA_2="-- Method 0:SMA, 1:EMA, 2:SMMA, 3:LWMA"; EXP_MA_3="-- Timeframe 0:current or minutes"; Use_MA1=false; MA1_Timeframe=0; MA1_Period=5; MA1_Method=0; MA1_Shift=0; Use_MA2=false; MA2_Timeframe=0; MA2_Period=0; MA2_Method=1; MA2_Shift=0; Use_MA3=false; MA3_Timeframe=0; MA3_Period=0; MA3_Method=1; MA3_Shift=0; Use_MA4=false; MA4_Timeframe=0; MA4_Period=0; MA4_Method=1; MA4_Shift=0; Use_MA5=false; MA5_Timeframe=0; MA5_Period=0; MA5_Method=0; MA5_Shift=0;

GBP/USD
SymbolGBPUSD (Great Britain Pound vs US Dollar) Period4 Hours (H4) 2009.06.29 00:00 - 2010.06.29 20:00 (2009.06.29 - 2010.06.30) ModelEvery tick (the most precise method based on all available least timeframes) ParametersiMagicNumber=532753; GMTOffset=2; TradeStartHour=1; TradeEndHour=24; Lots=1; TakeProfit1=20; TrailingStop1=140; TakeProfit2=1000; TrailingStop2=200; StopLoss=140; PendingPips=15; ExpireCandles=1; DelayOrder2=20; CheckSlope=false; DSS_name="Robby DSS Bressert Colored with alert v1.1"; DSS_EMA=8; DSS_stoch=13; Use_T3MA=false; T3MA_name="T3MA"; T3MA_Periods=400; T3MA_VolumeFactor=0.7; Channel_EMAPeriod=5; Channel_EMAShift=0; ShowComment=false; EXP_MA_1="-------- MA settings --------"; EXP_MA_2="-- Method 0:SMA, 1:EMA, 2:SMMA, 3:LWMA"; EXP_MA_3="-- Timeframe 0:current or minutes"; Use_MA1=false; MA1_Timeframe=0; MA1_Period=0; MA1_Method=1; MA1_Shift=0; Use_MA2=false; MA2_Timeframe=0; MA2_Period=0; MA2_Method=1; MA2_Shift=0; Use_MA3=false; MA3_Timeframe=0; MA3_Period=0; MA3_Method=1; MA3_Shift=0; Use_MA4=false; MA4_Timeframe=0; MA4_Period=0; MA4_Method=1; MA4_Shift=0; Use_MA5=false; MA5_Timeframe=0; MA5_Period=0; MA5_Method=0; MA5_Shift=0;
Attached File(s)
File Type: xls Jiva Dragon EA DSST3MA MAs v4.0.xls   31 KB | 458 downloads
  • #126
  • Jul 8, 2010 11:33am Jul 8, 2010 11:33am
Quoting Haishang
Jiva,

I tried to have same parameter as you do (see attached excel spreadsheet with all details), again not really succesful. Before I start testing my own parameters I try to gain common ground by making sure we are talking exactly about the same thing. That is the reason why i want to make sure that if we use same parameters we receive exactly the same results. Please note that I am on alpari uk live, that means my offset is GMT-2 in order to get same start time. is that correct and corresponds to your offset? I think so but for safety reasons...
Back testing is notorious for discrepancies among brokers due to the difference is data history. Even live trading can have huge differences if trades are executed or not, especially during fast moving markets. The differences in price quotes can be astounding, as you most likely are aware of.

Perhaps I can try to explain the dynamics I am using to test the parameters and you may try to adjust the settings accordingly on your tests that will come better inline with your brokers data history and live feed.

There is a dynamic among the parameters. Much like fine tuning a piano.
For this ea the basic players are: the stop, the limit, the pending distance and expiration time of the pending order. The main current is the channel setting and the others all dance around it.

In these 4h studies, the channel is very fast (2 or 3) and the candle expiration is only 1 candle. That is more or less remaining fixed for these studies. So it is trying to catch a breakout.

What can be varied with different settings (leaving the channel setting as it is) would be the limit, the stop, and the distance for the pending order.
Note:there is a setting called channel shift which I find useful in some applications, it will shift the channel according the number of candles specified. shift 2 would shift the channel 2 candles. It can improve the results for some pairs. The channel setting itself remains the same but it can be shifted.

From your above graph it appears that the ea is getting stopped out from the middle towards the end of the test. No doubt market conditions have changed and the stopping out is reflecting this change. Increased volatility is usually the culprit.

Open the chart. Study the trades, especially the losers. If the pending pips had been 20 instead of 15 would that have eliminated a lot of the bad entries? If so then change the setting to 20 pending pips and run the test again.

Again open the chart and study the losers. Trying to see where the dynamic is too much or too little. Play with the setting a bit. Would increasing the 1st profit make a difference, so that the 2nd position would not stop out at breakeven but would rather move on to catch a big move? But each time you make some change it should be just one change and then run the test again to see the results. If too many variables are changed then it may be too difficult to discern which change caused the result.

It is tedious no doubt, but one thing is: it is a sure way to learn inside and out the movements prices make during the trading. And how the dynamics work together. When the ea is set it plays like a baby grand. You can almost hear it.

It seems we are at the stage of practicing etudes and scales. But let us build some strength by the practice. And the proficiency is sure to come.

Please excuse the poetry, but I do not find this endeavor dry or mechanical in the least.

Good luck
  • #127
  • Jul 8, 2010 11:38am Jul 8, 2010 11:38am
Quoting jiva34
By using a trailing stop that is larger than the target it invalidates the trailing stop. That is the idea for these settings. No trailing stops are used. They are invalidated by the larger than target settings.

Once the 1st profit is reached the stop for the 2nd position moves to breakeven and fixed there. The idea is to let the 2nd position catch a big move if not it will stop out at break even....
Then why not use 0?

I have a friend that, when he is organizing his CD's and putting the contents on the label, if it's a blank CD, he will write on the label " Not used yet".
It's been a long 2 days!! Please excuse me and please don't be offended. It's important to understand sometimes.
  • #128
  • Jul 8, 2010 11:40am Jul 8, 2010 11:40am
Quoting sanyhuis
Then why not use 0?
I don't know. Haven't tried it. Thought it might clog the works, like the ea might pop a spring or something.
  • #129
  • Jul 8, 2010 11:44am Jul 8, 2010 11:44am
Quoting sanyhuis
Then why not use 0?

I have a friend that, when he is organizing his CD's and putting the contents on the label, if it's a blank CD, he will write on the label " Not used yet".
It's been a long 2 days!! Please excuse me and please don't be offended. It's important to understand sometimes.
  • #130
  • Jul 8, 2010 12:51pm Jul 8, 2010 12:51pm
Quoting jiva34

  • #131
  • Jul 8, 2010 7:33pm Jul 8, 2010 7:33pm
hi jiva, really liked your curve in post 107 but noticed that the modelling quality is n/a
so i tried using the exact same settings as yours, on data of higher modelling quality (with the gmt offset for my broker)

cheers..
________________________________________________________________

iMagicNumber=532753
GMTOffset=2
TradeStartHour=1
TradeEndHour=24
Lots=1.00000000
TakeProfit1=20
TrailingStop1=140
TakeProfit2=1000
TrailingStop2=200
StopLoss=140
PendingPips=15
ExpireCandles=1
DelayOrder2=20
CheckSlope=0
DSS_name=Robby DSS Bressert Colored with alert v1.1
DSS_EMA=8
DSS_stoch=13
Use_T3MA=0
T3MA_name=T3MA
T3MA_Periods=400
T3MA_VolumeFactor=0.70000000
Channel_EMAPeriod=5
Channel_EMAShift=0
ShowComment=1
EXP_MA_1=-------- MA settings --------
EXP_MA_3=-- Timeframe 0:current or minutes
Use_MA1=0
MA1_Timeframe=0
MA1_Period=5
MA1_Method=1
MA1_Shift=0
Use_MA2=0
MA2_Timeframe=0
MA2_Period=12
MA2_Method=1
MA2_Shift=0
Use_MA3=0
MA3_Timeframe=0
MA3_Period=20
MA3_Method=1
MA3_Shift=0
Use_MA4=0
MA4_Timeframe=0
MA4_Period=55
MA4_Method=1
MA4_Shift=0
Use_MA5=0
MA5_Timeframe=0
MA5_Period=197
MA5_Method=0
MA5_Shift=0
Attached Image(s) (click to enlarge)
Click to Enlarge Name: 2010-07-09_102256.gif Size: 10 KB
Click to Enlarge Name: 2010-07-09_102345.gif Size: 13 KB
  • #132
  • Jul 8, 2010 7:48pm Jul 8, 2010 7:48pm
sorry...
I modified trailing stop and pending pips.

v4.1
- no trailing stop when TrailingStop = 0
- take order immediately when PendingPips = 0
- some error messages changed when order failed

no other change
Attached File(s)
File Type: mq4 JIVA Dragon EA DSS T3MA MAs testv4.1.mq4   27 KB | 249 downloads
  • #133
  • Jul 8, 2010 10:05pm Jul 8, 2010 10:05pm
What happened to "EAs under construction III".....
  • #134
  • Jul 9, 2010 4:28am Jul 9, 2010 4:28am
Quoting reachjj
What happened to "EAs under construction III".....
I't's still there. I just looked at it.
  • #135
  • Jul 9, 2010 5:24am Jul 9, 2010 5:24am
Quoting sanyhuis
Alpari UK is +2
sorry, that's what I meant
  • #136
  • Jul 9, 2010 5:35am Jul 9, 2010 5:35am
Quoting jiva34
Back testing is notorious for discrepancies among brokers due to the difference is data history. Even live trading can have huge differences if trades are executed or not, especially during fast moving markets. The differences in price quotes can be astounding, as you most likely are aware of.

Perhaps I can try to explain the dynamics I am using to test the parameters and you may try to adjust the settings accordingly on your tests that will come better inline with your brokers data history and live feed.

There is a dynamic among the parameters....
Jiva, thank you so much for taking a lot of time to explain a bit further in detail how the ea is meant to work. This insight is very important for me to understand the corners from which you are coming by developing this wonderful tool.

And ultimately the more I understand about your mission the better i can put myself into useful help.

For a start, however, something most importantly strikes me. if you say that different broker's data make it basically impossible to compare each other findings (as you are on FXDD malta and I am on alari uk demo/live) even on the same settings.....how do we want to get into detail when we are both stuck and need to compare each others setting?

Perhaps i am too much interested into detail at this stage of developments and I should see only the broader picture by now. Is that what you want to tell me?
  • #137
  • Edited 3:17pm Jul 9, 2010 2:56pm | Edited 3:17pm
Quoting silv3r4sh
hi jiva, really liked your curve in post 107 but noticed that the modelling quality is n/a
so i tried using the exact same settings as yours, on data of higher modelling quality (with the gmt offset for my broker)

cheers..
________________________________________________________________

iMagicNumber=532753
GMTOffset=2
TradeStartHour=1
TradeEndHour=24
Lots=1.00000000
TakeProfit1=20
TrailingStop1=140
TakeProfit2=1000
TrailingStop2=200
StopLoss=140
PendingPips=15
ExpireCandles=1
DelayOrder2=20
CheckSlope=0
DSS_name=Robby DSS Bressert Colored with...
Thank you for taking the time and interest to look into the possibilities here.
I would like to see the details of the report if possible. The details I would like to see are found at the top of the report when it is saved by right clicking on the report. By using the copy and paste edit function they can be pasted into the post on ff. (rather than a screenshot)
The selection will look like Haishang's post #125. If it's possible for you to do that.
Also could you run a test using the higher modeling quality for the settings in post #106? And copy and paste the details as described above?
If you can that that would be great.
Thanks...
jiva.
  • #138
  • Jul 9, 2010 3:02pm Jul 9, 2010 3:02pm
Quoting Taiyakixz
sorry...
I modified trailing stop and pending pips.

v4.1
- no trailing stop when TrailingStop = 0
- take order immediately when PendingPips = 0
- some error messages changed when order failed

no other change
Dear friend, please don't ever say you are sorry about anything you are doing here. We are all indebted to you for your hard work and excellent writing.
Your gift is the miracle that turns a rough idea into a functioning reality!!
  • #139
  • Jul 9, 2010 3:34pm Jul 9, 2010 3:34pm
Quoting Haishang
Jiva, thank you so much for taking a lot of...
From experience with
Geoff's ea we can see that although there are differences there are enough similarities to make it workable. So it is not impossible.
Geoff's ea trades at a very calm time in the market place. Not usually any big sudden wild swings, in part that is why his ea is so successful for so many traders using different brokers.
This ea world is all new to me too.
So we are all learning together.
In a sense we are putting our heads together.
Let us see how to do this.
I am sure it is a matter of balance between details and dynamics. One without the other will not work.
There is a lot to be said for practice. Just taking the first steps and getting into motion, so even if it is in the wrong direction because there is movement then it can be turned to the right direction. Stand still and do nothing is hard to get going.
Mistakes are great teachers also.
So we are taking the first steps, and it is surprising the amount of success there is already.

A way to try to develop the ea (which is based on a specific strategy):
Set up the ea. Run a test using certain settings. Open the chart and study the trades. How would a slower or faster channel affect most of the trades, a higher or lower stop loss, a higher or lower time frame, etc. Would a longer candle expiration catch more good trades? More pips for a pending order stop bad entries? Adjust the settings and run the test again. Did that make a difference? Again study the trades.
So there is both need for details and dynamics.
At this point I know of no other way to proceed.
Also to keep an eye out if there is some improvement that can be made to the ea itself. Or maybe the ea is not closing trades as it should. The mechanics of it.
Please run some tests and try making adjustments after studying the trades on the chart. Let me know how it goes.
Thank you for taking an interest.
  • #140
  • Jul 9, 2010 10:13pm Jul 9, 2010 10:13pm
Quoting jiva34
From experience with
Geoff's ea we can see that although there are differences there are enough similarities to make it workable. So it is not impossible.
Geoff's ea trades at a very calm time in the market place. Not usually any big sudden wild swings, in part that is why his ea is so successful for so many traders using different brokers.
This ea world is all new to me too.
So we are all learning together.
In a sense we are putting our heads together.
Let us see how to do this.
I am sure it is a matter of balance between details and dynamics....
I think that accurate parameter tuning before finding basic strategy become just curvefitting...
I have read geoff's thread for long time, our approach is quite different from his ea.
He made ea because he wanted to automate his trade. He had already profitable strategy.
He modified little his strategy several times, but his basic strategy have not changed from the beginning.
In regard to jiva ea, it become a backtesting tool to look for strategy.
jiva, please look the strategy over again, good strategy is some profitable even with rough setting, then we should do parameter tuning to make it more profitable.
0 traders viewing now