Robot Lab Project | Page 43 | 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

  • #841
  • May 5, 2015 3:55pm May 5, 2015 3:55pm
Quoting abokwaik
{quote} MA Method used in settings is 1 (Exponential). Are you using the same on the chart ?
sorry my bad...It was simple MA on my chart...but nevertheless, it is still supposed to open a short. Changed it to exponential now, but it keeps opening a buy first order...
Attached Image (click to enlarge)
Click to Enlarge Name: NZDUSDM30.png Size: 22 KB
  • #842
  • May 12, 2015 1:16pm May 12, 2015 1:16pm
Quoting abokwaik
CLS_LWR_02 : Cut Losses Short - Let Winners Run Version 2 (with added SL, TP and Trailing Stop Inputs). To disable any of these inputs put it as large number (i.e. 99999). Testing results were better and draw downs are lower. Please note that this type of robots (which open many trades) is suitable only for higher time frames where spread effect is minimal. Again no robot is approved out of the Lab yet. You may only back test or forward test on demo. No live trading. Your input regarding this and previous...
Hi abokwaik,
let me congratulate you to your very fine work here first!
If wished I would like to contribute to developing the EAs. So I picked out this one first to study CLS and play with its features.
Doesn`t look too bad..... During the following days I hope to be able to take the time and analyse different TFs and pairs and perhaps find out optimations looking promising or additional functions the EA should have.
What do you think?
Keep on your excellent work.
Attached Image(s) (click to enlarge)
Click to Enlarge Name: CLS-H1-2.1.14-20.12.14-SL15.TS0.TP45.png Size: 19 KB
Click to Enlarge Name: CLS-H1-2.1.14-20.12.14-SL15.TS0.TP45.Report.png Size: 24 KB
Have a nice day
  • #843
  • May 12, 2015 2:11pm May 12, 2015 2:11pm
Quoting doerpn
{quote} Hi abokwaik, let me congratulate you to your very fine work here first! If wished I would like to contribute to developing the EAs. So I picked out this one first to study CLS and play with its features. Doesn`t look too bad..... During the following days I hope to be able to take the time and analyse different TFs and pairs and perhaps find out optimations looking promising or additional functions the EA should have. What do you think? Keep on your excellent work. {image} {image}
Hi, I also believe its an interesting EA. Nice work abokwaik. Good results doerpn, however sir, in your test you set the spread to 2. To my knowledge, in mt4, that refers to points, not pips...so to get "realistic" results, you should run with spread set to your avg +/- spread in points.
Best, Josh
  • #844
  • May 12, 2015 3:07pm May 12, 2015 3:07pm
Quoting JoshuaHope
{quote} Hi, I also believe its an interesting EA. Nice work abokwaik. Good results doerpn, however sir, in your test you set the spread to 2. To my knowledge, in mt4, that refers to points, not pips...so to get "realistic" results, you should run with spread set to your avg +/- spread in points. Best, Josh



Indeed

spread is in points so test with 20
Solutions ONLY!!!
  • #845
  • May 12, 2015 3:52pm May 12, 2015 3:52pm
Quoting BillYon
{quote} Indeed spread is in points so test with 20
ok, thanks for the two hints....will do it.
Have a nice day
  • #846
  • Edited 4:38am May 14, 2015 4:15am | Edited 4:38am
Quoting abokwaik
BASKET Project In order not to have things too complicated, here is the updated plan for the Basket Project: Create Basket versions of selected Robots: Create a Basket Trading Dashboard for manual trading I still have a concern regarding basket price calculation, so your thoughts will be valuable here: Suppose I have a USDBASKETthat contains 2 pairs (for simplicity) USDCHF, GBPUSD with Directional Weights +1,-1 respectively. To calculate Close Price (for example) for M1 basket candle I am currently using this formula : Close = (1*USDCHF.Close) +...
First of all thank you for sharing valuable information and codes.

I believe the bsket index should be created using geometric average. Using your example above, it would be:
BasketClose = ((1*USDCHF.Close) * (1/GBPUSD.Close)) ^ (1/N)

This way pip value is the exact pip value (0.0001), no need to convert or do additional calculations. JPY pairs should be divided by 100 during basket index calculations.

ps: I just realized that you have already mentioned Geometric mean in post 575. The sample you have given is using weights. I think to have real pip values it is better not to use any weights. If weights must be used than I believe Daily ATR of a long period should be used in an inverse fashion.
Example: (DayATR(90) of pair 1 * DayATR(90) of pair 2)^0.5 =CombinedAvgATR
Weight for pair 1 = CombinedAvgATR / DayATR(90) of pair 1
Weight for pair 2 = CombinedAvgATR / DayATR(90) of pair 2

This method would reduce the effect of the most volatile pair in basket, thus making the basket safer.
  • #847
  • May 14, 2015 6:00am May 14, 2015 6:00am
Quoting abokwaik
Correlation Divergence Trading (Two Pairs) I though about this, and reached the following: Lets take EUR/USD & GBP/USD as an example Calculate Change Percentage [ 100 x (Close-Open)/Open] on the desired time frame for both pairs, so we have EU_Change% and GU_Change% calculate Correlation Divergence CD = (EU_Change% - GU_Change%) x 100 Resulting CD values will oscillate below and above zero line, and can reach +100 or more and -100 or less Actions based on CD Values: +50 to +80 : (Moderate Signal) Buy GU and Sell EU : Close when CD becomes Negative...
Correlation Divergence is really interesting. Your original idea seems better, not the CCI method. I personally like your original idea better.
Since entry and exits are based on performances compared to open value, beginning of each day trades should be closed I guess.
Is that a correct assumption? Because all the figures will be hovering around 0 during the open of the day.
Please advise... Thanks and keep up the excellent work
  • #848
  • May 16, 2015 2:41pm May 16, 2015 2:41pm
Quoting GoldTheHun
{quote} First of all thank you for sharing valuable information and codes. I believe the bsket index should be created using geometric average. Using your example above, it would be: BasketClose = ((1*USDCHF.Close) * (1/GBPUSD.Close)) ^ (1/N) This way pip value is the exact pip value (0.0001), no need to convert or do additional calculations. JPY pairs should be divided by 100 during basket index calculations. ps: I just realized that you have already mentioned Geometric mean in post 575. The sample you have given is using weights. I think to have...
Hi GoldTheHun,

I moved the basket project to a separate thread Basket Robots. But haven't got the time lately to continue the project, Hopefully Soon.

Regards,

Khalil
No guts, no glory
  • #849
  • May 16, 2015 2:51pm May 16, 2015 2:51pm
Quoting GoldTheHun
{quote} Correlation Divergence is really interesting. Your original idea seems better, not the CCI method. I pershe onally like your original idea better. Since entry and exits are based on performances compared to open value, beginning of each day trades should be closed I guess. Is that a correct assumption? Because all the figures will be hovering around 0 during the open of the day. Please advise... Thanks and keep up the excellent work
Actually I didn't invest more on Correlation Divergence, but I have pending requests on making an MT4 version.

My schedule is very crowded these days, and the only way to implement all ideas is to quit my job, which I certainly will not do.

Or maybe I should make a career switch
No guts, no glory
  • #850
  • May 16, 2015 3:02pm May 16, 2015 3:02pm
Quoting doerpn
{quote} Hi abokwaik, let me congratulate you to your very fine work here first! If wished I would like to contribute to developing the EAs. So I picked out this one first to study CLS and play with its features. Doesn`t look too bad..... During the following days I hope to be able to take the time and analyse different TFs and pairs and perhaps find out optimations looking promising or additional functions the EA should have. What do you think? Keep on your excellent work. {image} {image}
Thanks doerpn,

Regarding the spread issue, if your broker is 5-digit (i.e. EURUSD has 5 decimals), you should use a spread of 20, While if your broker is 4-digit (i.e. EURUSD has 4 decimals) then you use 2 for the spread.

Note : CLS is designed for High time frames (H4 and above). If used on lower time frames it will have many consecutive losses which will be difficult to recover.
No guts, no glory
  • #851
  • May 17, 2015 9:59am May 17, 2015 9:59am
Quoting abokwaik
{quote} Thanks doerpn, Regarding the spread issue, if your broker is 5-digit (i.e. EURUSD has 5 decimals), you should use a spread of 20, While if your broker is 4-digit (i.e. EURUSD has 4 decimals) then you use 2 for the spread. Note : CLS is designed for High time frames (H4 and above). If used on lower time frames it will have many consecutive losses which will be difficult to recover.
Hi A.,
thanks for your quick answer. I tested EURUSD-Chart hour by hour with now spread of 20 (although with my broker I pay only 0.9 spread cost) to find the best settings.
Settings see screen please. Running from Jan14 to Mai15. It seems that this EA has potential. But I think we have to find out, why equity went down from Jan up to Sept. 14. There must have been a change in market behaviour since then.....

edit: sorry, the "Mai 15"-note on the graph-chart belongs to the end of the graph logically.
Attached Image(s) (click to enlarge)
Click to Enlarge Name: CLS-settings.png Size: 48 KB
Click to Enlarge Name: Graph-BT.CLS.png Size: 48 KB
Have a nice day
  • #852
  • Edited 1:05pm May 17, 2015 11:54am | Edited 1:05pm
Quoting doerpn
{quote} Hi A., thanks for your quick answer. I tested EURUSD-Chart hour by hour with now spread of 20 (although with my broker I pay only 0.9 spread cost) to find the best settings. Settings see screen please. Running from Jan14 to Mai15. It seems that this EA has potential. But I think we have to find out, why equity went down from Jan up to Sept. 14. There must have been a change in market behaviour since then..... edit: sorry, the "Mai 15"-note on the graph-chart belongs to the end of the graph logically. {image} {image}
Hi doerpn,

The 3 time frames used in your settings are D1, W1 & MN1.
Having a SL of 25 pips will generate large number of consecutive losses ---> large draw down.
Also using Trailing Stop of 35 pips will stop profits from running. ---> difficult recovery
My suggestion is to test it without SL and without Trailing Stop and increase Take Profit to at least 400 pips instead of 150.
As for closing time frame (Cut_Loss_TF) I would go for 2 (weekly).

Regards
No guts, no glory
  • #853
  • May 18, 2015 10:42am May 18, 2015 10:42am
Quoting abokwaik
{quote} Hi doerpn, The 3 time frames used in your settings are D1, W1 & MN1. Having a SL of 25 pips will generate large number of consecutive losses ---> large draw down. Also using Trailing Stop of 35 pips will stop profits from running. ---> difficult recovery My suggestion is to test it without SL and without Trailing Stop and increase Take Profit to at least 400 pips instead of 150. As for closing time frame (Cut_Loss_TF) I would go for 2 (weekly). Regards
Hi abokwaik,
I made BT as you recommended.....
Results see screens, please. Spread set to 20 pips. Ran it from 2.1.14 to 15.5.15....mmmhhhmmm.
Perhaps there should be the possibility to not open new trades if positions are in a loss or set a maximum for open trades....
Regards
Attached Image(s) (click to enlarge)
Click to Enlarge Name: CLS-settings-Daily.png Size: 44 KB
Click to Enlarge Name: CLS-EQ.png Size: 23 KB
Have a nice day
  • #854
  • May 18, 2015 3:32pm May 18, 2015 3:32pm
Quoting abokwaik
{quote} Actually I didn't invest more on Correlation Divergence, but I have pending requests on making an MT4 version. My schedule is very crowded these days, and the only way to implement all ideas is to quit my job, which I certainly will not do. Or maybe I should make a career switch
Thanks for the explanation. Well, whenever you have time, I will be waiting for it for sure.
Take care..
  • #855
  • Jun 4, 2015 11:44pm Jun 4, 2015 11:44pm
hi abokwaik,
i have sent pm to you.
regards
hukam
  • #856
  • Jun 6, 2015 8:38am Jun 6, 2015 8:38am
Quoting Motiram
Hello abokwaik, I see you are expert in Robots makings under this post also I saw you recap idea very elegantly under this link http://www.forexfactory.com/showthre...97#post8016897 Which make me impressed and finally giving an Open Source free EA to helping people for learning a coding your nature demonstrate you are a helping person i want to thank you for that Here I have my require to have one EA with the following requirements I will be grateful to you if you can help making the EA and can draw my request summarized as...
Hello abokwaik, when you get bit of time, please can you look at my request?

Thanks and Regards,
Moti
  • #857
  • Jun 24, 2015 5:20pm Jun 24, 2015 5:20pm
This is my first visit to the Robots Lab since over a month.

I have a new robot for you to try out.

the Cipher Robot.
Cipher means Zero, and its origin is Arabic from the word sifr.
So this robot trade the Cipher (or Zero) Levels.

I have provided many inputs that you can test, and find the most suitable combination for each pair.

Example :

If you set the Cipher Level to 100 pips (1000 points for 5-digit broker), the robot waits until the price comes near to 50 level, and opens pending orders (stop or limit as per the input) at 100 levels.

While if you set the level to 50 pips (500 points), the robot waits until the price comes near the 25(or 75) level, and opens pending orders on 50 and 100 level, and so on.

Minimum level allowed is 50 pips for 4-digit brokers and 500 points for 5-digit brokers.
Valid levels are multiples of 50 pips (or 500 points) --- (i.e: 50, 100, 150, 200 ....etc.)

These are two tests on EUR/USD 2015, all inputs are the same except one test is using multiple orders, while the other uses a single order.

Cipher with Multiple Open Orders

Attached Image (click to enlarge)
Click to Enlarge Name: cipher_MO.png Size: 54 KB


Cipher with Single Open Order

Attached Image (click to enlarge)
Click to Enlarge Name: cipher_SO.png Size: 58 KB



This is the settings file (provided as .txt) used in the above testing.

Attached File(s)
File Type: txt cipher_00.txt   < 1 KB | 640 downloads



.... Enjoy ....

Khalil
Attached File(s)
File Type: ex4 cipher_00.ex4   12 KB | 612 downloads
File Type: mq4 cipher_00.mq4   9 KB | 856 downloads
No guts, no glory
  • #858
  • Jun 24, 2015 5:29pm Jun 24, 2015 5:29pm
I have posted this question in another topic, but I think it is absolutely relevant over here.
Please tell me, what tpye of historical data do you use for backtesting?
Dukascopy, default MT4, or something else?
  • #859
  • Edited 7:46pm Jun 24, 2015 5:38pm | Edited 7:46pm
Quoting FxTechnic
I have posted this question in another topic, but I think it is absolutely relevant over here. Please tell me, what tpye of historical data do you use for backtesting? Dukascopy, default MT4, or something else?
Broker data... this one using FX.DD.
No guts, no glory
  • #860
  • Jun 25, 2015 6:24pm Jun 25, 2015 6:24pm
I don't know if anyone had coded this system :



Indicators:

 

  1. RSI(2) with the upper level at 90 and the lower level at 10
  2. SMA(200), and SMA(5).

Entry:

 

  1. Long Only When Price Closes Above SMA(200), AND Below SMA(5), With RSI(2) Below 10
  2. Short Only when Price Closes Below SMA(200), AND Above SMA(5), With RSI(2) Above 90


Exit:

 

  1. Exit Long when Price goes Above SMA(5)
  2. Exit Short when Price Goes Below SMA(5)

It looks interesting, taking short-term trend reversals in the direction of the long-term trend

What do you think ?

No guts, no glory
0 traders viewing now