EA31337 released for free | Page 3 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

No Free Lunch but all the Free Coffee you can drink 736 replies

SWAP free, rollover free Broker? 15 replies

Hi...best time to place order after news released 0 replies

Flyer! Version 2.00 is released. 0 replies

Cad News Released 1 Min Before... ! 43 replies

Attachments: EA31337 released for free
Exit Attachments

EA31337 released for free

  • #41
  • Apr 20, 2018 11:13am Apr 20, 2018 11:13am
Quoting shiva
{quote} Sorry for not being more clear. My thought was that, if the user sees a satisfactory running profit even before the preset targets are met, he/she could click a button to close all trades and retire for the day/week etc. This is what I did - Used an external "close all" script to close all open positions and disable the EA.
You can play with conditions, so like:

- when either account Equity is higher than 10% or 20%, or 'Max. daily balance > max. weekly', or 'Current day in profit', then 'close all profits' or/and 'suspend all strategies' (then another to unsuspend in the new week)

It gives some flexibility, it's a matter of testing and clarifying, which conditions or actions can be improved or added.
  • #42
  • Apr 20, 2018 11:14am Apr 20, 2018 11:14am
Quoting kenorb
{quote} You can play with conditions, so like: - when either account Equity is higher than 10% or 20%, or 'Max. daily balance > max. weekly', then 'close all profits' or/and 'suspend all strategies' (then another to unsuspend on the new week) It gives some flexibility, it's a matter of testing and clarifying, which conditions or actions can be improved or added.
Yes of course. Anyway, my idea is only cosmetic. Not functional. No worries.
  • #43
  • Apr 20, 2018 4:50pm Apr 20, 2018 4:50pm
Quoting pedrofx
This is a very complex and losing EA: http://en.zulutrade.com/trader/29357...0000&ref=57780 That's why it's free... {image}
i never thought that losing is complex
  • #44
  • Edited 9:54pm Apr 20, 2018 7:40pm | Edited 9:54pm
Quoting llxx77dd
{quote} i never thought that losing is complex
Main EA is made of 3 versions (4th is on the way). The one on Zulu was experimental Rider version, not Lite. Secondly, as already mentioned in post #12, Zulu couldn't handle hundreds of trades per day and 80% of trade updates were rejected causing different results on their side.

Would you like to learn what EA in Rider mode can do? My backtests showed on broker data, starting as low as $1000 or $2000 deposit to over $100000 profit in first 6 months, then few million in a year, even on a high spread. It's a risky mode, but very powerful. Data accurate and directly from the broker.

Here are the different backtest stats using different configurations and it was back in 2015 (v1.062) before hundreds of other improvements:

Summary backtest log:
All [2015.01.01-2015.06.30 based on MT4 FXCM backtest data, spread 20, 9,5mln ticks]
- profit, trades, profit factor, e, DD$, DD%, params
* Deposit: £2000 (default, spread 20)
- £129847.70 5406 1.25 24.02 83328.01 72.67% TrendMethodAction=17 (d: £2k, spread 20, Rider, auto)
- £84538.74 5282 1.21 16.01 64259.13 72.67% TrendMethodAction=16 (d: £2k, spread 20, Rider, auto)
- £71144.36 5323 1.29 13.37 34450.52 70.68% TrendMethodAction=25 (d: £2k, spread 20, Rider, auto)
- £77576.22 5258 1.44 14.75 24957.91 77.13% Market_Condition_12=9 Action_On_Condition_12=6 (Rider, auto)
- £31634.22 5503 1.25 5.75 16191.08 67.12% Market_Condition_7=6 (Rider, auto)
* Deposit: £1000 (default, spread 20)
- £7242.02 23883 1.11 0.30 4013.08 52.91% TradeWithTrend=1 (Lite)
- £1719.22 32541 1.04 0.05 1914.97 55.15% TradeWithTrend=0 (Lite)
- £8505.73 5452 1.18 1.56 6310.19 79.89% TradeWithTrend=0 (Rider, auto)
- £5345.28 4571 1.09 1.17 15073.88 83.51% TradeWithTrend=1 (Rider, auto)

I'm also attaching details of one of the test.
Attached File(s)
File Type: txt EA31337 v1.062 backtest stats.txt   6 KB | 431 downloads
  • #45
  • Apr 20, 2018 11:05pm Apr 20, 2018 11:05pm
Any hints of how to optimize a stratgy that overall loses?

Eg. GBPJPY 2014-2016:

Profit factor: 1.34, Total net profit: 55.27EUR [+217.46/-162.19] (2876.73pips), Total orders: 548 (Won: 67.2% [368] / Loss: 32.8% [180]) - CCI M15
Profit factor: 1.32, Total net profit: 26.15EUR [+109.16/-83.01] (1444.02pips), Total orders: 60 (Won: 63.3% [38] / Loss: 36.7% [22]) - Envelopes M30
Profit factor: 1.11, Total net profit: 185.85EUR [+1952.74/-1766.89] (25831.79pips), Total orders: 2068 (Won: 77.8% [1609] / Loss: 22.2% [459]) - WPR M15
Profit factor: 1.06, Total net profit: 105.27EUR [+1966.44/-1861.17] (26013.05pips), Total orders: 1674 (Won: 71.5% [1197] / Loss: 28.5% [477]) - WPR M30

Total net profit: -203.68 EUR

Seems MT4 optimizer doesn't yield any results during optimization so maybe it has to be done by hand.
  • #46
  • Apr 20, 2018 11:41pm Apr 20, 2018 11:41pm
Please include one version of EA and one set of files that are profitable. Without that thread it makes no sense.
Please excuse the bad English via Google Translate.
  • #47
  • Apr 21, 2018 1:17am Apr 21, 2018 1:17am
Quoting kenorb
{quote} Main EA is made of 3 versions (4th is on the way). The one on Zulu was experimental Rider version, not Lite. Secondly, as already mentioned in post #12, Zulu couldn't handle hundreds of trades per day and 80% of trade updates were rejected causing different results on their side. Would you like to learn what EA in Rider mode can do? My backtests showed on broker data, starting as low as $1000 or $2000 deposit to over $100000 profit in first 6 months, then few million in a year, even on a high spread. It's a risky mode, but very powerful....
Hi Kenorb, thank you for sharing your system.
Do you have in Zulu or Myfxbook or elsewhere an audited forward test?
In Zulu you have an experimental version, ok. Where we can see the results of the good version?
Thanks.
  • #48
  • Apr 21, 2018 2:32am Apr 21, 2018 2:32am
Quoting pedrofx
{quote} Hi Kenorb, thank you for sharing your system. Do you have in Zulu or Myfxbook or elsewhere an audited forward test? In Zulu you have an experimental version, ok. Where we can see the results of the good version? Thanks.
Please re-read posts #12 and #44 about Zulu, the service is not compatible due to their limitations and it's not going to work as I've already tried. For forward results they were shared in this thread, check post #36.

I didn't run any forward tests, during constant development. EA needs at least few weeks to warm-up, so when next version is released, then I'd end-up maintaining multiple versions, and I don't have time for this. I'm a Mac user for 20+ years and it's not going to change, and running EA under macOS/Linux (such as wine) is troublesome for me and platform is really not compatible with my everyday stack which I'm using. Once I find some time, I'll publish.
  • #49
  • Apr 21, 2018 2:49am Apr 21, 2018 2:49am
Hi Kenorb,I've downloaded the source for v1.076, but not able to compile. See the attached
Attached File(s)
File Type: txt errorrs.txt   6 KB | 524 downloads
  • #50
  • Edited 3:10am Apr 21, 2018 2:51am | Edited 3:10am
Quoting chrisdk2017
Any hints of how to optimize a stratgy that overall loses? Eg. GBPJPY 2014-2016: Profit factor: 1.34, Total net profit: 55.27EUR [+217.46/-162.19] (2876.73pips), Total orders: 548 (Won: 67.2% [368] / Loss: 32.8% [180]) - CCI M15 Profit factor: 1.32, Total net profit: 26.15EUR [+109.16/-83.01] (1444.02pips), Total orders: 60 (Won: 63.3% [38] / Loss: 36.7% [22]) - Envelopes M30 Profit factor: 1.11, Total net profit: 185.85EUR [+1952.74/-1766.89] (25831.79pips), Total orders: 2068 (Won: 77.8% [1609] / Loss: 22.2% [459]) - WPR M15 Profit factor: 1.06,...
You posted all results of all profitable strategies. That's good for you. So if you found at least 4 profitable strategies (based on 3 indicators CCI, Envelopes and WPR and 2 timeframes M15 and M30). So why not to disable the other losing one, and focus your optimization process first on these 3.

Optimization is complex and time-consuming, not really designed for humans. Since you've got a good starting point, focus on profitable CCI, Envelopes and WPR-based strategies, make them better by tweaking/optimizing only 1-2 parameters at the same time (with only 2 timeframes: M15 and M30 which worked for you). You may consider running optimization only for one strategy at a time, by disabling the other one. During optimization, consider disabling boosting (to have more reliable results). After optimization of one strategy, test them together (the selected one which are good). After that find another good strategy candidate. Focus optimization on one strategy and make it good, then re-enable all selected one. Repeat the process. After you found group of strategies which meet your needs, final optimization stage is to optimize non-based strategy params (such as boosting).
  • #51
  • Edited 3:23am Apr 21, 2018 3:03am | Edited 3:23am
Quoting shiva
Hi Kenorb,I've downloaded the source for v1.076, but not able to compile. See the attached {file}
You're missing the include files. So you need to solve the dependencies. Make sure you're using multiple repositories across includes which are created for the same version. Classes from the other repo are also progressing, so not every version of classes are compatible with the main code (classes repo is not versioned using the same method, but main repo has specific submodule references to it, which points to the compatible version). Check the Compilation guide. To see which version is compatible with which files, on the main repo switch to the specific tag and browse the files (next to the folder you should see git's SHA1 hash). On macOS I do symlinks across repo which was cloned recursively, so everything is consistent, on Windows, good luck.
  • #52
  • Apr 21, 2018 3:56am Apr 21, 2018 3:56am
Quoting kenorb
{quote} Please re-read posts #12 and #44 about Zulu, the service is not compatible due to their limitations and it's not going to work as I've already tried. For forward results they were shared in this thread, check post #36. I didn't run any forward tests, during constant development. EA needs at least few weeks to warm-up, so when next version is released, then I'd end-up maintaining multiple versions, and I don't have time for this. I'm a Mac user for 20+ years and it's not going to change, and running EA under macOS/Linux (such as wine) is...
I run mt4 under Linux/wine.

The prefix32 directory is 1.5 GB so I could upload it to Mega, if you are interested. Then you only need to install the newest wine and then run mt4.
  • #53
  • Apr 21, 2018 3:58am Apr 21, 2018 3:58am
Quoting kenorb
{quote} You posted all results of all profitable strategies. That's good for you. So if you found at least 4 profitable strategies (based on 3 indicators CCI, Envelopes and WPR and 2 timeframes M15 and M30). So why not to disable the other losing one, and focus your optimization process first on these 3. Optimization is complex and time-consuming, not really designed for humans. Since you've got a good starting point, focus on profitable CCI, Envelopes and WPR-based strategies, make them better by tweaking/optimizing only 1-2 parameters at the same...
Thanks,

I just wonder why the strategies are shown as profitable but overall net profit is negative?

So difficult to know where the source of loss is.
  • #54
  • Apr 21, 2018 4:25am Apr 21, 2018 4:25am
Quoting kenorb
{quote} Please re-read posts #12 and #44 about Zulu, the service is not compatible due to their limitations and it's not going to work as I've already tried. For forward results they were shared in this thread, check post #36. I didn't run any forward tests, during constant development. EA needs at least few weeks to warm-up, so when next version is released, then I'd end-up maintaining multiple versions, and I don't have time for this. I'm a Mac user for 20+ years and it's not going to change, and running EA under macOS/Linux (such as wine) is...
The forward test published in post #36 is for only 1 day (20/Apr). Do you have a more consistent forward results? Live or demo?
  • #55
  • Apr 21, 2018 4:26am Apr 21, 2018 4:26am
Quoting chrisdk2017
{quote} Thanks, I just wonder why the strategies are shown as profitable but overall net profit is negative? So difficult to know where the source of loss is.
Probably you've copied it from the Journal log which may drop some line. Check for the text file which was generated in the tester/files folder, the summary of profit/loss should match. Consider posting it.
  • #56
  • Apr 21, 2018 4:32am Apr 21, 2018 4:32am
Quoting pedrofx
{quote} The forward test published in post #36 is for only 1 day (20/Apr). Do you have a more consistent forward results? Live or demo?
Wait then for others to post it then, as I don't have anything up-to-date for the reason mentioned for few times already. Keeping it maintained and babysit not-stable platform takes time and I'm not using the same tools as others do.
  • #57
  • Apr 21, 2018 4:42am Apr 21, 2018 4:42am
Quoting kenorb
{quote} Probably you've copied it from the Journal log which may drop some line. Check for the text file which was generated in the tester/files folder, the summary of profit/loss should match. Consider posting it.
Attached.
Attached File(s)
File Type: txt GBPJPY-1000EUR-2014.01.01-2015.12.31-3spread-(5)-M30-report.txt   45 KB | 329 downloads
  • #58
  • Edited 5:31am Apr 21, 2018 4:53am | Edited 5:31am
Quoting chrisdk2017
{quote} I run mt4 under Linux/wine. The prefix32 directory is 1.5 GB so I could upload it to Mega, if you are interested. Then you only need to install the newest wine and then run mt4.
I've got 20+ MT4 platforms and few MT5 on my macOS, so running platform it-self under wine is not the problem. Availability is not the problem either, as I've got hundreds of platforms since build 200 (even some MT3 files) which I can run under wine. It's the question how bad it is. The versions are not consistent and easy to be broken. Due to the complexity of the code, the EA may not compile (as per post #27) on one version, but works for newest version which comes with MetaTrader 5, because MT4 is no longer maintained and cannot be downloaded from official sources anymore, so it's a bit unmaintainable mess. What is worse, most of the brokers don't provide up-to-date version of the platform. Another example, the older code needs to be compiled with older compilers, newer with newer versions, sometimes specific part of the code needs to be compiled with the specific build, or needs to be refactored to babysit the MQL language changes. Of course, I've reported hundreds of bugs to MT, most of the bugs never been fixed, either for no reason, or reason that MT4 is no longer supported, or they simply don't care, or they do these problems deliberately to make it difficult for people to create something fully working. Another thing which breaks, the wizard, not every wine version can handle the installer (like this issue), so you end up with fixing and hacking more libraries to make it work, and at the end, next version pop up and you start again. Another thing is how to perform the proper backtest, half of the time you're doing reverse-engineering of the formats which were deliberately created this way, to make the trouble for you, than focusing on the actual backtesting. So instead of doing proper backtesting, you need to hack into many file formats such as FXT (now HCC), HST, SRV (where some broker values are hardcoded), some of them are heavily encrypted, in order to modify the values to simply run the backtest exactly how you want it (otherwise the result aren't reliable). You think it's easy? I can pay few hundred bounty if you write a Python script to read and modify SRV file (especially a lot size value), good luck. And finally, MT4 is no longer supported and going to be dropped soon. Furthermore, MQL deliberately removed MT4 installer last year from their site, including command-line compiler which I was using, and compiling files using MetaEditor is a nightmare (like the exit codes are not consistent, and they clearly said they're not going to fix it). So trust me, running under wine for such project is a huge problem. Therefore, this is not the platform which I want to work with.
  • #59
  • Apr 21, 2018 5:02am Apr 21, 2018 5:02am
I understand.

You could port your ea to QuantConnect lean, a c# trading engine.

https://github.com/QuantConnect/Lean

Mono works fine on Linux. Probably also on Mac.
  • #60
  • Apr 21, 2018 5:58am Apr 21, 2018 5:58am
Quoting kenorb
{quote} You're missing the include files. So you need to solve the dependencies. Make sure you're using multiple repositories across includes which are created for the same version. Classes from the other repo are also progressing, so not every version of classes are compatible with the main code (classes repo is not versioned using the same method, but main repo has specific submodule references to it, which points to the compatible version). Check the Compilation guide. To see which...
Hi Kenorb, I have all the files now (Hopefully) most other errors have disappeared but these remain.

MA_Period_Fast' - variable already defined ea-input.mqh
S_ALLIGATOR' - undeclared identifier EA31337.mq4
S_BANDS' - undeclared identifier EA31337.mq4
0 traders viewing now