Robot Lab Project | Page 55 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Financial Markets Observatory Lab 1,095 replies

Indicators Lab 236 replies

Rel's Lab 103 replies

Nikola's Trading Lab 247 replies

Spud's Lab Accident 22 replies

Attachments: Robot Lab Project
Exit Attachments

Robot Lab Project

  • #1,081
  • Mar 7, 2016 3:36am Mar 7, 2016 3:36am
Quoting roughtrader
{quote} Hi abokwaik! I would like to see a larger sample of data, 30 trades is hardly enough to determine if this is good or not, I suggest at least 200 trades. if it is any good we should see a up sloping equity curve by then, I tested ideas that was great for about 50 trades going straight up, but then straight down after that. just saying, a larger sample data is needed to see if it will continue look like that.
I completely agree.

The problem is : I am really a busy man. The time I allocate to system coding is actually stolen from the time of important things (work, family, sleep , recreation, ... etc).
I post these systems so that people can test and report their findings, because I really don't have the time to test and optimize, and I also don't have a good quality data to rely on.
No guts, no glory
  • #1,082
  • Mar 7, 2016 4:04am Mar 7, 2016 4:04am
Quoting Pipologist
{quote} Was wondering if anyone has used or tested in Jan/Feb 2016.
This is a test on GU/EU M30 for 2016 (with large number of trades due to short time frame)

Attached Image (click to enlarge)
Click to Enlarge Name: cd_02_EU_GU_M30_2016.PNG Size: 55 KB


and This is a test on GU/EU M15 for 2016 (with large number of trades due to short time frame)

Attached Image (click to enlarge)
Click to Enlarge Name: cd_02_EU_GU_M15_2016.PNG Size: 55 KB
No guts, no glory
  • #1,083
  • Mar 7, 2016 4:16am Mar 7, 2016 4:16am
Quoting abokwaik
{quote} This is a test on GU/EU M30 for 2016 (with large number of trades due to short time frame) {image} and This is a test on GU/EU M15 for 2016 (with large number of trades due to short time frame) {image}
Those settings as indicated in the images for your initial test runs?
  • #1,084
  • Mar 7, 2016 4:47am Mar 7, 2016 4:47am
Quoting Pipologist
{quote} Those settings as indicated in the images for your initial test runs?
No, initial tests were on Daily time frame. Those tests are on lower time frames, so settings are different, and produce larger number of trades as roughtrader requested.
No guts, no glory
  • #1,085
  • Mar 7, 2016 4:50am Mar 7, 2016 4:50am
Any way, settings were never optimized on Correlation Divergence EA.

Also the method used for detecting divergence was only one idea.
I am sure there are better ways to detect divergence than cci indicator.
No guts, no glory
  • #1,086
  • Mar 7, 2016 4:57am Mar 7, 2016 4:57am
I am starting a short period forward test for Corr_Diver_02 on EU/GU M15, on MT5.
It was not possible to create a Trade Explorer as FF doesn't support MT5.

Here is a screen shot ..., and I will post results ... as frequent as I can.

Attached Image (click to enlarge)
Click to Enlarge Name: cd_02_ft.PNG Size: 129 KB
No guts, no glory
  • #1,087
  • Mar 7, 2016 5:55am Mar 7, 2016 5:55am
Quoting abokwaik
I am starting a short period forward test for Corr_Diver_02 on EU/GU M15, on MT5. It was not possible to create a Trade Explorer as FF doesn't support MT5. Here is a screen shot ..., and I will post results ... as frequent as I can. {image}
Khalil, are you going to make MT4 version?
  • #1,088
  • Mar 7, 2016 6:05am Mar 7, 2016 6:05am
[Quote = abokwaik; 8795829] Qu te parece? Hay que hacer una versin MT4 de la misma, o no? [/ Quote]


Abokwaik Hello !
Of course! I'm already waiting MT4 version !
Thanks again for your efforts friend!
  • #1,089
  • Mar 7, 2016 6:17am Mar 7, 2016 6:17am
Hopefully before end of week I will make an MT4 version of Correlation Divergence (2 pairs), using multiple Divergence Detection methods.
No guts, no glory
  • #1,090
  • Edited 7:36am Mar 7, 2016 7:25am | Edited 7:36am
I have been running back tests on breakouts for the last 5 days with a generic breakout EA really loving EA's

Let me know what you think of the results i would love to filter some losses but i can't with my EA i'm finding quite steady results so far anyway

I'm focusing on 1:4 as I have heard EA's results vary when scalping?

Happy to share

First one is 2 years

2nd and 3rd screenshot last 4 months

Catching the London open trends
Attached Image(s) (click to enlarge)
Click to Enlarge Name: Screenshot4.png Size: 111 KB
Click to Enlarge Name: Screenshot1.png Size: 112 KB
Click to Enlarge Name: Screenshot2.png Size: 106 KB
  • #1,091
  • Mar 7, 2016 8:26am Mar 7, 2016 8:26am
Quoting Cizz
I have been running back tests on breakouts for the last 5 days with a generic breakout EA really loving EA's Let me know what you think of the results i would love to filter some losses but i can't with my EA i'm finding quite steady results so far anyway I'm focusing on 1:4 as I have heard EA's results vary when scalping? Happy to share First one is 2 years 2nd and 3rd screenshot last 4 months Catching the London open trends {image} {image} {image}
It looks OK to me. Good Job.
No guts, no glory
  • #1,092
  • Mar 7, 2016 9:46am Mar 7, 2016 9:46am
Hello Abo,

I am stil running TwoMAoneRsi with still good results.see my trade explorer (CSS Alpha 1).

Nevertheless Could you add some filters:

-Minimum distance price/short MA
-Maximum distance price/long MA

Any other idea??

THanks for the great work.

S
  • #1,093
  • Mar 7, 2016 9:58am Mar 7, 2016 9:58am
abokwaik,

Thank you for sharing your ideas. I've been reading your thread and decided to try one of your Robots MA Cross WT 02.
I have only started since LO and had 8 trades fired off. Sometimes 2 trades at once fired off when only 1 should have.
Closed
2 losses (minimal loss)
1 57.9 Profit

Open
5 Open in Profit 267.4 Profit

Entry & Exit Method "0"
30 pip Entry Buffer
Attached Image (click to enlarge)
Click to Enlarge Name: 3ma.JPG Size: 197 KB


MTPockets
  • #1,094
  • Mar 7, 2016 10:08am Mar 7, 2016 10:08am
Quoting sbenichou1
Hello Abo, I am stil running TwoMAoneRsi with still good results.see my trade explorer (CSS Alpha 1). Nevertheless Could you add some filters: -Minimum distance price/short MA -Maximum distance price/long MA Any other idea?? THanks for the great work. S
I thought I did that already! must be my age, if not.
I will check and update in Elite thread.
No guts, no glory
  • #1,095
  • Mar 7, 2016 10:11am Mar 7, 2016 10:11am
Quoting MTPockets
abokwaik, Thank you for sharing your ideas. I've been reading your thread and decided to try one of your Robots MA Cross WT 02. I have only started since LO and had 8 trades fired off. Sometimes 2 trades at once fired off when only 1 should have. Closed 2 losses (minimal loss) 1 57.9 Profit Open 5 Open in Profit 267.4 Profit Entry & Exit Method "0" 30 pip Entry Buffer {image} MTPockets
I never used it on M5, but it looks good so far.
No guts, no glory
  • #1,096
  • Mar 7, 2016 10:32am Mar 7, 2016 10:32am
Quoting abokwaik
{quote} I thought I did that already! must be my age, if not. I will check and update in Elite thread.
ooops sorry....
I missed your answer , i have just seen it.

I will check this thanks a lot.

S
  • #1,097
  • Mar 10, 2016 5:09am Mar 10, 2016 5:09am
Quoting abokwaik
Correlation Divergence Trading (Two Pairs) Third Version Corr_Diver_03 (MT5 Only) Updates : Correlation Factor : 1 for Positive, -1 for Negative Correlation between the 2 pairs Pairs Ratio : 1st pair to 2nd pair volatility ratio (better entered manually than calculationg using ATR specially on lower time frames) CCI Extereme Value : If using CCI period 2, leave this to 0 (zero), but when using periods like 20, 30 ... etc, you can use higher values for this parameter (i.e. 100). This parameter will filter the trades, the higher value the less trades....
Hi Abo,
I am testing the Corr_Diver_03 in Alpari with strategy tester it works well. But when I test it with exact inputs and conditions in FxPro, no trade at all. Trying other inputs still no trade too. Any tips ?

Thank you, great job anyway. I like the idea of correlation divergence.
  • #1,098
  • Mar 10, 2016 5:23am Mar 10, 2016 5:23am
Quoting HeWee
{quote} Hi Abo, I am testing the Corr_Diver_03 in Alpari with strategy tester it works well. But when I test it with exact inputs and conditions in FxPro, no trade at all. Trying other inputs still no trade too. Any tips ? Thank you, great job anyway. I like the idea of correlation divergence.
It's fine on ActivTrades.

Are your receiving any errors in the log ?
No guts, no glory
  • #1,099
  • Mar 10, 2016 7:03am Mar 10, 2016 7:03am
Quoting abokwaik
{quote} It's fine on ActivTrades. Are your receiving any errors in the log ?
Hmm no errors in the log. Just no trade. I had deleted the history folder too but no luck. Maybe because the "ECN" account, not sure. I attach the journal log maybe it's useful. Anyway other brokers work fine, hehe.
Attached File(s)
File Type: txt journal log.txt   33 KB | 452 downloads
  • #1,100
  • Edited 9:28am Mar 10, 2016 8:51am | Edited 9:28am
Quoting abokwaik
Correlation Divergence Trading (Two Pairs) Third Version Corr_Diver_03 (MT5 Only) Updates : Correlation Factor : 1 for Positive, -1 for Negative Correlation between the 2 pairs Pairs Ratio : 1st pair to 2nd pair volatility ratio (better entered manually than calculationg using ATR specially on lower time frames) CCI Extereme Value : If using CCI period 2, leave this to 0 (zero), but when using periods like 20, 30 ... etc, you can use higher values for this parameter (i.e. 100). This parameter will filter the trades, the higher value the less trades....
Hi Abo,
I observe in the code (can read yet unable to modify ) that there are only two condition for closing open positions, those are:

  1. Profit_EQ_Perc_Inc or Loss_EQ_Perc_Dec has been reached.
  2. TP or SL has been hit.

By relying to those 2 conditions, we are hard to optimize the input value. Cause besides optimizing the CCI period and CCI Extreme Value, we also need to find the match value between Profit_EQ_Perc_Inc, Loss_EQ_Perc_Dec, and lot size. We may find that trade will get very long or very short to close due to mismatch lot size or the percentage. The problems will occurs more when we move to other timeframes. Also as our account grows/shrinks, we need to constantly adjust the lot size, I think it is not backtest-able and not forwadtest practice.

Just wondering why not stick to your original ideas to close the positions when CD turns back into opposite sign (positive / negative)

Attached Image


By having this original condition, we only need to optimize 2 inputs: CCI period and CCI Extreme Value regardless timeframe we use.

So in term of optimization -> (2 parameters applicable for any timefrime) vs (5 parameters multiplied by the number of timeframe).

And also, why do you disable the "automatic Pair_Ratio" feature ? (maybe you have a consideration)
0 traders viewing now