Welcome, Guest. Please login or register.

May 28, 2020, 02:29:26 PM

Login with username, password and session length

Shoutbox

Recent

Members
Stats
  • Total Posts: 752
  • Total Topics: 68
  • Online Today: 23
  • Online Ever: 144
  • (August 14, 2019, 12:20:12 PM)
Users Online
Users: 0
Guests: 11
Total: 11

11 Guests, 0 Users

Author Topic: SFE Stealth EA  (Read 14262 times)

Attractor

  • Administrator
  • Sr. Member
  • *****
  • Posts: 298
    • View Profile
    • SmartForexExpert
Re: SFE Stealth EA
« Reply #60 on: February 28, 2020, 11:20:14 PM »
yes I'm very conscious about the over optimization, don't worry, I always try to search essential rules with sense, obviously sometimes I can mistake, but I know.
I'm working now on stealth, I have a lot of  things to check in all the EAs after this fool week.
I will inform of the advances.

Attractor

  • Administrator
  • Sr. Member
  • *****
  • Posts: 298
    • View Profile
    • SmartForexExpert
Re: SFE Stealth EA
« Reply #61 on: March 02, 2020, 01:17:41 AM »
I have been working without rest since the bad experience last week...
suffering in livel last week movements in live, gave me some good ideas :)

The new stealth will be online soon, with very very big improvements for be protected against more strength conditions of the market if this kind of market comes again last happened last days.

An advance of the new version, I want to finish and publish it this week.







- timeframe m15 added on v 1.2 removed, is not enough reliable in some market conditions






Byte

  • Jr. Member
  • **
  • Posts: 15
    • View Profile
Re: SFE Stealth EA
« Reply #62 on: March 04, 2020, 04:19:36 PM »
Thank you for the continued improvements. This looks very impressive

Attractor

  • Administrator
  • Sr. Member
  • *****
  • Posts: 298
    • View Profile
    • SmartForexExpert
Re: SFE Stealth EA
« Reply #63 on: March 04, 2020, 04:54:33 PM »
Thank you for the continued improvements. This looks very impressive

you are welcome. The new version 2.0 is finished and I attached to my official signal already, I'm only waiting to see the first new trades to publish it in mql5.
Meanwhile, in the first message of this thread, find the final bactkest and the information of the new version.

jasper2603

  • Jr. Member
  • **
  • Posts: 4
    • View Profile
Re: SFE Stealth EA
« Reply #64 on: March 04, 2020, 05:07:10 PM »
Thanks for the update and the backtest, but 78% history quality.. Could you do an mt4 backtest with 99% history quality? And with fixed lotsize

Attractor

  • Administrator
  • Sr. Member
  • *****
  • Posts: 298
    • View Profile
    • SmartForexExpert
Re: SFE Stealth EA
« Reply #65 on: March 04, 2020, 05:22:15 PM »
Thanks for the update and the backtest, but 78% history quality.. Could you do an mt4 backtest with 99% history quality? And with fixed lotsize

This is the quality when I use 'every tick' and last 35 minutes in my laptop. For more qualtity, I have to use 'every tick based on realt ticks' but at least my computer (or metatrader) can't process this backtest, because we are using 19 symbols from 2008, is a lot of information.

At this moment this EA doesn't open trades in the 'zero' hour, then, don't backtest with real ticks is less relevant than the previous version.

And backtest with 'real ticks' from 2008 (if all this info exist)  is also a problem because fortunately, the spreads right now are a lot better than years ago.

After the EA is published, if someone handle to backtest in real ticks, will be welcome  :)

Castellers de Vilafranca

  • Jr. Member
  • **
  • Posts: 70
    • View Profile
Re: SFE Stealth EA
« Reply #66 on: March 04, 2020, 09:18:39 PM »
Joel I am confused. This backtest should be done in 0.1 lots like all of the other backtests.

There is absolutely no way to compare this to previous versions thus no way to be able to know the actual drawdown risks.

You must have a normal 0.1 lot backtest version, right?

Castellers de Vilafranca

  • Jr. Member
  • **
  • Posts: 70
    • View Profile
Re: SFE Stealth EA
« Reply #67 on: March 04, 2020, 09:24:29 PM »
Thanks for the update and the backtest, but 78% history quality.. Could you do an mt4 backtest with 99% history quality? And with fixed lotsize

I 100% agree, this needs to be in fixed lot size!!!!!!

And 78% needs to be improved for sure!! I don't care if it takes 35 hours......

Attractor

  • Administrator
  • Sr. Member
  • *****
  • Posts: 298
    • View Profile
    • SmartForexExpert
Re: SFE Stealth EA
« Reply #68 on: March 05, 2020, 11:36:16 AM »
version 2.0 just published. Please, check the first message of the thread, some details are there.
The update has moved from 1.X to 2.X because this is a major update, I mean, 2 important features has been added:

1. new autolot mode:  risk %/ trade
2. the 'Volatility Safe Mode' parameter

1. I was not much follower of the risk %/trade method for set the size of the trades because in my first EAs, mainly price action methods, sometimes happens that more violent trades have more successful profit ratios, then, if you do an homogeneus % risk or % win based on this criteria, you can lost some profability at final. Maybe this is this way in some of my EAs, I will see, but the truth is that applied to EAs like strealth, I like a lot the results, because large movements or less large movements, all have the same risk, the results are less rollercoaster, and have another good property that is that knowing the max. number of trades the EA can open (posted in the first message), we can calculate the max % of loss if the EA opens MAX orders and all hit the SL. (the sl of the order, is more far than the real sl, is for avoid spikes in the spread).
Then, since now and while I don't change my mind, I will use this autolot criteria in my signal and my recommendations of risk will be based in this criteria.
For this reason I posted 3 backtest with 3 different risk / trade, for have an idea of the max. historical DD in % in relation to each risk.

2. This is a very important parameter. If anyone backtest the EA since 2008 with this parameter in 0 (disabled), will see bad results in the years of major volatility after the 2008 crisis. Then, this filter, remove setups if the chart give signals of big movement. We know this kind of filter is very difficult to get something good, but seems I have found something that works quite well as we can see in the backtest, changing 1 to 0, 0 to 1 and see the differences.
Obviously, in years like 2017 and 2018, with very low movements in the whole time,have this filter enabled, can decrease the profitability.
Then, is on your side decide or not use this filter. In my accounts and for any user that want an 'all terrain' EA that also can survive if things turns very volatile, I recommend left this parameter enabled.

--------------------------

Then, for setup right now:
- decide if left the volatility parameter in 1.
- left the value of the new risk%/trade or use another method of lot size. (look the backtest on the first message for have an idea of the historical DD, you can see the max. DD of the balance or the equity in %)
- check if your broker have all the symbols, because I added two 'rare' symbols in this version, for example XAUEUR and USDSGD.

** the m15 timframe has been ripped, all the trades are based on the m5 bars, like the version 1.0,1.1.

Last, maybe I expressed bad, my computer don't do nothing when I try the backtest with 'ticks based on real ticks'. But believe me, I don't need always backtest with this precision for know when I'm getting improvements or not, the main question in the backtesting is the overoptimization. Use 1 minute bars, or every tick, or real ticks, will be more or less important depending of the kind of EA. Right now, this EA offers enough guarantee for me using every tick. and while I develop I use 1 minute bar, I only go do more detailed backtest for the fina presentation, but with 1 minute I have a enough close ideo of the performance.

And for different kind of backtest with fixed lot, etc. As the mt5 version is published, everyone can do his backtest and share here if you want. Remember to use the Test_Full_List=1 to test all the symbols of the list at the same time.

Pesonally I don't want to spend much time in presentations of different kind of backtest because my value here is improving the EAs. After the lessons of last days, stealth has been the first, I also finished a breakout version 1.1 (basically for avoid re-entry in long movements) , and I'm with NS now.

** I'm changing the information panels with a more lightweight panel, for improve the process speed and can have several EAs in the same panel.
Looks something like this:



If there is any error, the panel will write an error message.

« Last Edit: March 05, 2020, 11:44:59 AM by Attractor »

Castellers de Vilafranca

  • Jr. Member
  • **
  • Posts: 70
    • View Profile
Re: SFE Stealth EA
« Reply #69 on: March 05, 2020, 09:16:24 PM »
Joel is synchronization still happening? (I just cut and paste below in case I am using the wrong word)

Parameter: Synchronization ID

This parameter identifies configuration profiles and synchronizes through the Smartforexexpert website. Put 0 in this parameter if you don't want to use this feature.

For use this feature, you have to add the address:   https://smartforexexpert.com to the list of 'allow web request'

Castellers de Vilafranca

  • Jr. Member
  • **
  • Posts: 70
    • View Profile
Re: SFE Stealth EA
« Reply #70 on: March 05, 2020, 09:20:15 PM »
Also Joel I checked in the backtest folder. There is no XAUEUR symbol in the backtests.

Castellers de Vilafranca

  • Jr. Member
  • **
  • Posts: 70
    • View Profile
Re: SFE Stealth EA
« Reply #71 on: March 06, 2020, 11:57:09 AM »
Also another thing Joel is there is now no option for Order Comment.

I like to add comments to help search for everything on FX Blue or myfxbook plus I don't like brokers seeing what EA's I am using so I always change comment. The new Stealth does not have this option unfortunately.

petersurrey

  • Jr. Member
  • **
  • Posts: 11
    • View Profile
Re: SFE Stealth EA
« Reply #72 on: March 07, 2020, 08:48:31 AM »
Thanks for the updates Attractor - I note that almost all of your profit in your previous SFE stealth EA version(s) has come from one pair - XAUUSD - ( custom analysis of your myfxbook account without this pair). Is there any reason why your backtests fail to reflect this.. giving it about the same profit or worse than many others pairs.. assuming you are assigning the correct $/pip value? I much prefer an EA that targets specific high probability trades, rather than a more scattergun approach almost always on the market - perhaps your next version could reflect this?

petersurrey

  • Jr. Member
  • **
  • Posts: 11
    • View Profile
Re: SFE Stealth EA
« Reply #73 on: March 07, 2020, 08:59:08 AM »
Can you also confirm whether there has been any change to entry/exit logic in your latest version?

petersurrey

  • Jr. Member
  • **
  • Posts: 11
    • View Profile
Re: SFE Stealth EA
« Reply #74 on: March 07, 2020, 10:40:11 AM »
Joel can you attach the back test for version 2 to your first post as they are still registering version 1 - thanks:)