EA31337 released for free | Page 18 | 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

  • #341
  • May 29, 2018 3:42pm May 29, 2018 3:42pm
Quoting zeramlakireg
Hot damn! Look at the smoking green pips. {image} 8,000 pips in 4 days! Surreal!
Lol, crazy. Did you change any settings? You trade also manually or leave it full auto?
Anything above the line is an uptrend, below the line is a downtrend.
  • #342
  • May 29, 2018 3:47pm May 29, 2018 3:47pm
Quoting Degolep
{quote} Lol, crazy. Did you change any settings? You trade also manually or leave it full auto?
It's running on full auto, but will probably look for an EA to close all position every Friday as I'll be changing my pairs every week.
  • #343
  • May 29, 2018 4:23pm May 29, 2018 4:23pm
Quoting kenorb
{quote} You can lower RiskMargin param to 0.5 (0.5% per trade), alternatively RiskRatio to 0.5. Also, try to lower BoostTrendFactor param to 1.1 or 1.2. Another way is to disable Boosting section. These changes should generate smaller lot sizes.
Appreciate it chief
  • #344
  • May 29, 2018 4:31pm May 29, 2018 4:31pm
Quoting zeramlakireg
{quote} Yeah, Im also just holding back from going live until I have a few months worth of data. Will probably test first on a live micro. I'm using the default settings on my multicurrency TE but reduced the riskmargin = 0.1 per pair so it trades lower lots. Cheers!

Sure , ganna test the lot size settings on a another demo and see how it goes
  • #345
  • Edited 6:14pm May 29, 2018 5:15pm | Edited 6:14pm
Quoting ImpAsh
{quote} Hi, here is it: {file}

I see everyone is posting nice results but when I do back test it blows up in the dirt. I can't figure out what am I doing wrong. I used the Set file shared by Impash. Below is 1m and 15m test on EU since January 2018.

Settings I used are same posted by impash
Attached Image(s) (click to enlarge)
Click to Enlarge Name: 8ad8838b324af94b6698c496fa792f66.png Size: 22 KB
Click to Enlarge Name: 57ae51076a9d5aeb09d8b75d5ff7f512.png Size: 21 KB
Game taught me the game. And it didnt spare me the rod while teaching
  • #346
  • May 29, 2018 5:17pm May 29, 2018 5:17pm
Quoting zeramlakireg
Hot damn! Look at the smoking green pips. {image} 8,000 pips in 4 days! Surreal!

Do you mind sharing your set file please.
Game taught me the game. And it didnt spare me the rod while teaching
  • #347
  • Edited 6:18pm May 29, 2018 5:43pm | Edited 6:18pm
Quoting Singh1
{quote} I see everyone is posting nice results but when I do back test it blows up in the dirt. I can't figure out what am I doing wrong.{image} {image}
In my opinion, all 90/99% backtesting (where data is converted using old unsupported scripts) provides just fake and unreliable results (used commonly by people who're not aware of wrongdoing or by the scammers) and I believe most of the EAs are backtested in the wrong way, but the majority believes it's the right way, that's why all of them struggling to be profitable despite showing the perfect curves. In one screenshot you've got 22 mismatched charts errors, these are the data errors (this should not happen), and since you've set the files to read-only (I assume), the platform doesn't have the ability to correct the errors. What you see in nice green bar - 90%, it is the fake value hardcoded by the script in FXT file (that's why you was asked to set the file as read-only, I believe). If you drop the read-only flag, most likely the files would be wiped out by the platform and marked as the corrupted and platform simply won't run the test. Same for the screenshot with n/a (platform is confused by the invalid value which you're trying to force). These all forward-results just showing it's most likely the truth. For more explanation, check post #22. Basically, I'm using completely different scripts for optimization and backtesting on which I'm relying on (which now I'm convinced even more, that my method is more reliable than the common FXT-read-only aka "99%" modelling method). Check the steps described on 'Backtesting in MT4' page which I've written which uses the more reliable method (which uses ready-made FXT/HST files or use the Python scripts). I'll soon share the easier method for backtesting.
  • #348
  • May 29, 2018 6:15pm May 29, 2018 6:15pm
Quoting kenorb
{quote} In my opinion, all 90/99% backtesting (where data is converted using old unsupported scripts) provides just fake and unreliable results (used commonly by the scammers) and I believe most of the EAs are backtested in the wrong way, but the majority believes it's the right way, that's why all of them struggling to be profitable despite showing the perfect curves. In one screenshot you've got 22 mismatched charts errors, these are the data errors (this should not happen), and since you've set the files to read-only (I assume), the platform...
So in short I should forward test and nope I didnt select any read only option but okay will test forward on demo. Thanks
Game taught me the game. And it didnt spare me the rod while teaching
  • #349
  • May 29, 2018 6:30pm May 29, 2018 6:30pm
One question, should I set TP to auto or there is a recommended TP and SL?
Game taught me the game. And it didnt spare me the rod while teaching
  • #350
  • Edited 7:17pm May 29, 2018 7:03pm | Edited 7:17pm
Quoting Singh1
One question, should I set TP to auto or there is a recommended TP and SL?
The EA is using different approach for TP/SL comparing to other EAs which is by using dynamic indicator values per strategy-basis (e.g., see: post #258), so there is no auto, or single TP/SL param to change, but multiple places. TP/SL are set in form of trailing stops per strategy. The default settings are the recommended one unless you can optimize the better one. To change TP/SL, you need to find the strategy of your choice in the list of params (grouped by sections), then adjust it. If you need to set the fixed one, there is Fixed value as well (but it's not recommended). So there are no auto values for TP/SL, just default settings based on the optimization results.
  • #351
  • May 29, 2018 9:45pm May 29, 2018 9:45pm
Quoting Singh1
{quote} Do you mind sharing your set file please.
I've already shared the set file for EURUSD only TE. Just look it up in the attahments (paperclip icon on top of the page).

The mql market has a lot of EAs doing 1,000% daily in backtests but will blow your account in a couple of days (even hours)!

So, don't trust your backtest/optimization if you are not kenorb. XD

Just forward test the EA.

Also, don't expect other account's performance to be replicated in yours. You will start forward test in a different market condition and your account history (where the EA bases its lot sizing) are different.

Cheers!
  • #352
  • May 29, 2018 10:20pm May 29, 2018 10:20pm
Quoting zeramlakireg
{quote} I've already shared the set file for EURUSD only TE. Just look it up in the attahments (paperclip icon on top of the page). The mql market has a lot of EAs doing 1,000% daily in backtests but will blow your account in a couple of days (even hours)! So, don't trust your backtest/optimization if you are not kenorb. XD Just forward test the EA. Also, don't expect other account's performance to be replicated in yours. You will start forward test in a different market condition and your account history (where the EA bases its lot sizing) are...

Yeah I am not expecting same results right of the back but I did not expect this big of a open loss and it was even bigger than that. I think I am gonna wait until someone finishes proper forward testing and author makes necessary changes after feed back. For now this EA is not ready for live account and others are doing forward testing already so I will just wait.
Attached Image(s) (click to enlarge)
Click to Enlarge Name: 5d7c3b874cc4c0a46741dea4720d7e4b.png Size: 88 KB
Click to Enlarge Name: 1819777922a31b27bc331f1cf0610a68.png Size: 78 KB
Game taught me the game. And it didnt spare me the rod while teaching
  • #353
  • Edited 12:41am May 30, 2018 12:24am | Edited 12:41am
Quoting Singh1
{quote} Yeah I am not expecting same results right of the back but I did not expect this big of a open loss and it was even bigger than that. I think I am gonna wait until someone finishes proper forward testing and author makes necessary changes after feed back. For now this EA is not ready for live account and others are doing forward testing already so I will just wait. {image} {image}
Personally, I forward test an EA to be familiar with how it trades. Especially how it opens the lot size in relation to my account size. We have different brokers, leverages, capital, risk tolerance, etc.

IMO, if you have not forward tested an EA in demo using your prefered broker, leverage, and capital which you plan to use live, don't even go live with any EA.

Edit:
Regarding backtesting ver 1.072, I was not able to find any range of date where the EA was profitable in my backtests. It always blew up the account. I used 90% data quality with no mismatches. Yet, the result of the TE is drastically different.
  • #354
  • May 30, 2018 1:32am May 30, 2018 1:32am
be carefull with this.....
Attached Image (click to enlarge)
Click to Enlarge Name: Sin título.png Size: 106 KB
  • #355
  • May 30, 2018 2:00am May 30, 2018 2:00am
Quoting Singh1
{quote} Yeah I am not expecting same results right of the back but I did not expect this big of a open loss and it was even bigger than that. I think I am gonna wait until someone finishes proper forward testing and author makes necessary changes after feed back. For now this EA is not ready for live account and others are doing forward testing already so I will just wait. {image} {image}
I don't think you can get a low open dd with this ea due to how it trades many strategies at the same time.

So yes, it is risky, big reward means big risk, which so far cannot be avoided.

If you want low dd means you need single trade eas IMHO.
  • #356
  • May 30, 2018 5:07am May 30, 2018 5:07am
the most curious thing is that the EA continues to open sell positions
Attached Image (click to enlarge)
Click to Enlarge Name: Sin título.png Size: 111 KB
  • #357
  • May 30, 2018 5:27am May 30, 2018 5:27am
Quoting arkan1976
the most curious thing is that the EA continues to open sell positions {image}
Which Set file are you using?
  • #358
  • May 30, 2018 6:15am May 30, 2018 6:15am
Quoting arkan1976
the most curious thing is that the EA continues to open sell positions {image}
Your lot size is way too big.

Set it to 0.01 for per 1000 USD / EUR.

The problem is that the initial lot size calculation is using the account leverage.

Maybe also turn off boosting.
  • #359
  • May 30, 2018 6:48am May 30, 2018 6:48am
[quote=chrisdk2017;11114684]{quote} Your lot size is way too big. Set it to 0.01 for per 1000 USD / EUR. The problem is that the initial lot size calculation is using the account leverage. Maybe also turn off boosting.[/quot
thanks for response!
  • #360
  • Edited 11:56am May 30, 2018 7:13am | Edited 11:56am
I think people are too much focusing on the old version, keep running the same default configuration over and over again without changing anything, then they see the first week to be very profitable (then other people are panicking to follow the same), then the next week goes wrong and everybody is blaming EA. Unfortunately, the v1.072 with its default settings isn't balanced out, and it's not suitable for live and it was already mentioned (at least not with the default params). If you'd like to avoid high lot sizes, try at least turning off the boosting, then try again.
There were several fixes improved in the further versions, it may not give you 100% a week, but it aims to be more balanced.

For example, can anybody test the next version, which is the v1.073? It trades over 10 times less often than v1.072, but the logic should be very similar. Also, v1.073 was optimized and tested on the larger dataset (2014-2016, instead of 2014-2015). I'm attaching the backtesting results to compare v1.072 vs v1.073.

----

EA31337-Lite-Backtest-v1.072
Symbol EURUSD (Euro vs US Dollar)
Period 30 Minutes (M30) 2014.01.06 00:00 - 2015.12.30 23:30 (2014.01.01 - 2015.12.31)
Bars in test 24856 Ticks modelled 154351318 Modelling quality 89.64%
Initial deposit 10000.00 Spread 20
Total net profit 18930.69 Gross profit 91018.87 Gross loss -72088.18
Profit factor 1.26 Expected payoff 3.25
Absolute drawdown 1619.57 Maximal drawdown 9401.33 (30.06%) Relative drawdown 30.06% (9401.33)
Total trades 5829 Short positions (won %) 4460 (48.68%) Long positions (won %) 1369 (42.00%)
Profit trades (% of total) 2746 (47.11%) Loss trades (% of total) 3083 (52.89%)

Attached Image (click to enlarge)
Click to Enlarge Name: EURUSD-v1.072-10000USD-2014-2015year-20spread-M30-DS-test.gif Size: 18 KB


----

EA31337-Lite-Backtest-v1.073
Symbol EURUSD (Euro vs US Dollar)
Period 30 Minutes (M30) 2014.01.06 00:00 - 2016.12.30 21:30 (2014.01.01 - 2016.12.31)
Bars in test 36856 Ticks modelled 358224900 Modelling quality 89.76%
Initial deposit 10000.00 Spread 20
Total net profit 4310.23 Gross profit 34124.90 Gross loss -29814.67
Profit factor 1.14 Expected payoff 6.71
Absolute drawdown 2126.29 Maximal drawdown 4601.64 (36.89%) Relative drawdown 36.89% (4601.64)
Total trades 642 Short positions (won %) 351 (40.17%) Long positions (won %) 291 (42.27%)
Profit trades (% of total) 264 (41.12%) Loss trades (% of total) 378 (58.88%)


Attached Image (click to enlarge)
Click to Enlarge Name: EURUSD-v1.073-10000USD-2014-2016year-20spread-M30-DS-test.gif Size: 19 KB

Note re above chart: Before you say there is a drawdown in the middle, it is not exactly, as the equity graph rocked up (the ticker line) and the EA could potentially hit the higher balance, but it didn't close on time.

Here is another chart for v1.073 with spread 10:
Attached Image (click to enlarge)
Click to Enlarge Name: EURUSD-10000USD-2014-2016year-10spread-M30-DS-test.gif Size: 19 KB


----

Few notes:
- v1.072 got max 9401USD (30% DD) in the drawdown in 2 years, and v1.073 got only 4601USD in DD in 3 years, which is two times less (so safer to use).
- v1.073 trades 13 times less per year (642/3y=214 vs 5829/2y=2914.5), so that means you don't waste your money into broker's fees/spreads or/and commissions.
- Expected payoff is 2 times higher for v1.073 (6.71 vs 3.25).
- Please note that above 2 charts are for spread 20, which is very high (most of EAs are tested with 10 or less). If you've got lesser spreads with your brokers, profits can potentially double, as you pay less to the broker.
Attached File(s)
File Type: txt EURUSD-v1.073-10000USD-2014-2016year-20spread-M30-DS-test.txt   8 KB | 317 downloads
File Type: txt EURUSD-v1.072-10000USD-2014-2015year-20spread-M30-DS-test.txt   8 KB | 254 downloads
0 traders viewing now