Simplicity your Life
- #2
- Dec 31, 2006 1:01pm Dec 31, 2006 1:01pm
- Joined Feb 2006 | Status: Blah blah blah | 1,410 Posts
Looks similar to mine 
The breaking of a wave cannot explain the whole sea.
- #4
- Dec 31, 2006 1:55pm Dec 31, 2006 1:55pm
- Joined Feb 2006 | Status: Blah blah blah | 1,410 Posts
The breaking of a wave cannot explain the whole sea.
- #10
- Jan 1, 2007 7:22am Jan 1, 2007 7:22am
- Joined Dec 2005 | Status: Com Member = Scammer | 5,408 Posts
- #11
- Edited 8:32am Jan 1, 2007 7:30am | Edited 8:32am
Your backtest statement is rubbish... Everyone knows MT4 backtesting is useless trash.
If so, please give us as many arguments as possible:
1. MT4 backtesting is useless trash, because:
-
-
-
2. Backtesting could be reilable, if:
-
-
-
Thanks in advance, I'm counting on you!
Quidquid latine dictum sit altum viditur
- #12
- Jan 1, 2007 2:28pm Jan 1, 2007 2:28pm
- Joined Feb 2006 | Status: Blah blah blah | 1,410 Posts
The breaking of a wave cannot explain the whole sea.
- #13
- Jan 1, 2007 2:29pm Jan 1, 2007 2:29pm
- Joined Feb 2006 | Status: Blah blah blah | 1,410 Posts
The breaking of a wave cannot explain the whole sea.
- #14
- Jan 1, 2007 6:55pm Jan 1, 2007 6:55pm
Oh, we have an expert here...
The most important is this one:
Do you use Indicator that use Close price?
or use the Close[0] to retrieve info or you try to get the Higher timeframe...
If so your are totally wrong....
In testing, the Close does not return a good value, but in real time it does (On a new candle you only have the open, not the close).
So by exemple: if you use RSI to make your decision in your backtest, it will not show the good value because it is base on PRICE_CLOSE (It average the value it suppose to have) ... And you know why now!
So i was not able to backtest my EA because of this issue.
Sorry for my bad english translation 
Oh, we have an expert here...
If so, please give us as many arguments as possible:
1. MT4 backtesting is useless trash, because:
-
-
-
2. Backtesting could be reilable, if:
-
-
-
Thanks in advance, I'm counting on you!
Cant completely agree that backtesting is useless trash, but its definitely unreliable.
- #16
- Edited 2:25am Jan 2, 2007 1:22am | Edited 2:25am
Be careful of what you say!, Hes right! Backtest is just there to help you, not confirm your 100% right! And yes i know what i'm talking about....
many people are just posing to be experts in something or thy are just jelous
I like to check if they are know something - almost never... so i like to tease em a bit ... but to the point - essentially yes and moreover curve fitting is a well known technique in physics and mathematics so past results only indicate but NEVER guarantee future performance. The second of many isues is a quality of data as garbage in garbage out, for example there is a problem with Alpari's who, as I've heard changed (manipulated) data in their database changing platforms.
thats why backtest can only indicate which one EA, and which parameters should be checked live... and then you can check additionally if a backtest on new data is in line with live. the second technique is to backtest only on a part of a database and "forward test" on the rest - most recent, but this is also never 100% reliable....
Thanks for your input
Quidquid latine dictum sit altum viditur
- #18
- Jan 2, 2007 11:33am Jan 2, 2007 11:33am
- Joined Feb 2006 | Status: Blah blah blah | 1,410 Posts
The breaking of a wave cannot explain the whole sea.
Hi dottore:
I appreciate the depth of your contributions to this board! Please go on (english isn't my primary language either!).
In your last post I picked the term "curve fitting". Could you please explain this? I have read that term a couple of times now, relating it to trading, but I don't understand what you mean with "curve fitting".
I have searched www.answers.com and Googled, but the articles all apply to scientific topics.....not trading.
Thanks for any help.
Peter.
<o
></o><o
></o>I’ll give you an example. You are trying to find a best moving average as a support for a candles in D1. Suppose that you begin with a SMA50, but on the chart you see that too many candles are penetrating it so you change period to 55. Then you change SMA to WMA and the result is better. This process is called curve fitting in FX – you are trying to find the best curve to your purposes.<o> Per analogy curve fitting in EA means to test different parameters to find the best most profitable (in tha past )) ) combination. </o
>In creating and optimizing EA’s this process is rather dangerous and often leads to failure – we call it over-optimization. Such systems are generally good only over short period in time and then collapse. I believe a good analogy can be a F1 bolid, unbeatable in very specific conditions but useless and dangerous in country roads. <o
></o>I’m pretty sure that EA above is an example of this kind of system, although I have two or three arguments against; first – the system doesn’t rely on incidental events but frequently occurring retracements, second – earlier versions of it were good. The main argument against it is a size of data – 100 “transactions” that’s not much from the statistical point of view. So we decided to test it live as demo supplies slightly different data. 7000% a year (in next 100 trades) is impossible but if it will be 10-15x capital that will be ok.. J <o
></o><o
> </o>Happy New Year to you! <o
></o><o
> </o>))<o></o> Quidquid latine dictum sit altum viditur