molanis.com/forum - Tools for MetaTrader : The place to ask about the best expert advisor builder, expert advisor downloads, and expert advisor programming

Hi there hardworking pple!


The good thing I want to say about Molanis SB is that for those who has 99% determination and 1% inspiration then this software is definately for you.


Without coding and it's ease of use the 99% determination and the 1% inspiration you have in you the SPEED between downloading that Trading Idea in your head to the Trading Blocks takes just mere minutes.


What is even better is that as your EA starts to take shape the perfectionist in you will get to try out every new alteration settings in just seconds.


I'll be a millionaire soon. You will too!!!!!!!!!!


http://www.youtube.com/watch?v=1rtLDYCgJ-A&feature=plcp


Google me at kitz3281.

Agree with you on the software for sure.

Regarding your backtest, beware of 63.83% drawdown. That is the thing with martingale EA's--the risk of ruin must be considered.

Thanks RJo!



REDUCING DRAWDOWNS.
Thats why my backtest always begin with just $100. This puts my EA in a very tight spot and in order to survive a one year test (average 13,000 bars, for M30) my EA must not cause a DRAWDOWN that burns the account.



Another way is to just maintain the MaxNoOfPosition at 1. Developers may have the impression that to make money the MaxNoOfPosition must be set as many as you can (100) so that when your Trading Logic is true you will profit a lot. But if your EA has a Trade Accuracy of only 75% then the other 25% and with the MaxNoOfLots set at 100 will surely whack the account when your EA meets a DRAWDOWN. Setting MaxNoOfPosition at 1 will avoid this.
Be contented with MaxNoOfPosition at 1 and focus on improving Trade Accuracy please. Setting MaxNoOfLots at 100 does not provide the shortcut to the profits you are looking for. Improving on Trade Accuracy does.



Lastly when my EA passes the one year test it will then be subjected to a ten year test. The years it profits and the years it does not profit are all noted down (compile on Excel). I do this with all my EAs.



Next my EAs are all compared side by side for 'matching'. I want to see if they can cancel each other out. Means if one EA does not profit for the year 2008 then I'll get one that profits in 2008 to make them cancel each other out. This matching is done with the two EAs for all ten years. Such when one has a DRAWDOWN with losses then the other will pull it up, neutralize that DRAWDOWN with it's profits.



Next if this two EAs still has a combined 'lost year', say in 2003, then I'll get in the third EA that can profit in 2003 to neutralize that 'lost year' with profits in 2003 from the third EA (It will get harder and harder to match after this point).



So next if this three EA still has a 'lost year' ... bla bla bla, bla bla .. I'll get in the fourth EA to ...... you get the picture.



Repeat the 'matching' process until you pass through ten years without losses.



As good as I am I still need so far at least seven EAs to achieve a 'no lost' report for ten years straight through. Seven EAs re-filed as one huge EA. I just need one EA to trade at least 50 trades per year. That may seem small but seven EAs means I get around 350 trades a year already. Each with a high PROFIT TRADE and high CONSECUTIVE WINS and high PROFIT FACTOR the 'matching' result will be - an EA that trades 350 times per year with averagedly high PROFIT TRADE, high CONSECUTIVE WINS & high PROFIT FACTOR.




So for all reading this you will know what you must do if you want your EA to trade 700 times a year - repeat twice the entire above mentioned process!

You almost stole the record for the longest post by a non-administrator from cfx... almost.

Oh really.

Hehehe

Interesting approach you just need to be sure to check that mt4 reaches 90% or more in modeling quality. http://www.google.com/search?q=How+to+A ... =firefox-a

Hi Guys, nice of you to drop in ... !!!!!!!!!!

I've just visited the link you gave me. Just browsed through though, didn't really absorb what's it all about. But I'll be back there for sure.

It sounds like '90% Modelling Quality' is PURE historical data? as compared to what we normally have for testing?


Its the 'Wind Tunnel' for serious EA developers???? ))))))

if you do less than 90% your backtesting is not as accurate so results may not be valid.

Hey, this 90% thing sounds familiar: http://www.molanis.com/forum/viewtopic.php?p=9506#p9506

But seriously, is modeling quality relevant when kit is using open prices? From my limited understanding of how ST works, modeling quality measures intra-bar data. Of course this assumes kit is trading on completed bars. Or are using open prices in ST always a bad idea?

I agree with you 100%.


Firstly, I dont understand why one should create an EA for the M1 timeframe.
I have been in love with forex for more than 20 years and really understands that we traders really need to live, eat, sleep, shift (sports mode ), play and spend time with our family relatives and friends. Maybe one has learnt that trading 24hours isn't living anymore so he wants his 1minute EA to take over.


I have read many Technical Analysis books and the works of our great pioneers such as Walter Bressert and the likes. Their illustration time frames ARE Days, Weeks, Months and even Years! And I am sure when Dr. George C Lane invented the Stochastics 60 years ago and when Gerard Appel histogrammed the MacD 40 years ago they never would expect that we traders are testing their formula's on minute frames! Yes the technology we have was not available to their time yet but I am sure some things do not change - that Fundamentals and Cycles drives pip movements in their hundreds and thousands, not 10 pips. What happened to the Basics? I am making M30 EAs and I am already feeling the guilt!!!


Once our EAs are already working does re-testing them on Grade1 hostorical data make any difference compared to testing the EAs on the Hostorical Data that comes standard with your MT4? And that its a Multi-Timeframe EA dependent on higher timeframes and one that uses open prices?


Price moves as slow as Whales when only one market is open and they move as fast as Sharks when Germany, London and New York opens simultaneously, in my case at GMT+8 (Singapore). How will your scalping EA survive at these times?


So my take on 90% Quality Historical Data is that unless ur really good at creating EAs for 10-20 pips then this issue is not really a concern isn't it?


But if you succed I want to know about it. I may be entirely wrong!!!!