Backtesting not the same as in the championship

Back to topics list To post a new topic, please log in or register

BlackThorn 2006.10.06 22:00

Renat

I posted the question that I do not get the same results as the server for the competition.

I am not out to prove that MetaTrader does not work, I think that it is great software and very well written. I also understand that sometimes isolated bugs can occur.

I am just concerned that I cannot get sashken's EA to work. I use MIG's data feed and apart from that their should not be any differences.

You must understand that it concerns me that I cannot get the same result because I spend lots of time with the backtester and optimizer to test my strategy, but it is all a waste of time if I cannot trust the results. I am not expecting exactly the same results using different data but it must be within about 10% of each other.

This is what happens when I test the EA from 2006/10/01 to 2006/10/07 using a daily period(GPD/USD):
1 2006.10.06 00:00 buy 1 4.00 1.8790 0.0000 1.8930 0.00 10000. 00
2 2006.10.06 23:59 close at stop 1 4. 00 1. 8711 0.0000 1.8930 -3160. 00 6840. 00

Here is exactly the same run but with the time between 2006/10/01 to 2006/10/08
1 2006.10.06 00:00 buy 1 4.00 1.8790 0.0000 1.8930 0.00 10000. 00
2 2006.10.06 23:59 close at stop 1 4. 00 1. 8708 0.0000 1.8930 -3280. 00 6720. 00

Here is exactly the same run but with the time between 2006/10/01 to 2006/10/30
1 2006.10.06 00:00 buy 1 4.00 1.8790 0.0000 1.8930 0.00 10000. 00
2 2006.10.06 23:59 close at stop 1 4. 00 1. 8705 0.0000 1.8930 -3400. 00 6600. 00

When I put the time between 2006/10/01 to 2006/10/06 no trade occur.

In my opinion this is already worrying - how can a future date change the results. It also states that it is closing out at 23:59 tonight?

What is also strange is the fact the every time I click start, the program collects data and downloads about 300kb

What do you think is possible problems?

Thanks

Fallacies, Part 1: Money Management is Secondary and Not Very Important

The first demonstration of testing results of a strategy based on 0.1 lot is becoming a standard de facto in the Forum. Having received "not so bad" from professionals, a beginner sees that "0.1" testing brings rather modest results and decides to introduce an aggressive money management thinking that positive mathematic expectation automatically provides positive results. Let's see what results can be achieved. Together with that we will try to construct several artificial balance graphs that are very instructive.

1515	DxdCn 2006.10.07 19:05 MIG's data feed ????

582

irusoh1 2006.10.07 19:47

I think the key here is where you start the first trade. His trade started at 1. 8750. So he was lucky to hit his take profit target.
When I ran it on my tester. It started at 1.8762 for some reason and failed to hit the target. The result was some -2500 loss because t/p target was not hit. I lowered t/p traget to 120 and got 2 similar trades as in championship. So these 8 pips really made a difference between 1st place and bust. I am not sure this strategy will hold water in the long run.
I am not sure why tester starts from different point. You might run a visual mode and take a closer look.

6	BlackThorn 2006.10.08 19:36 DxdCn wrote: MIG's data feed ???? I use http://www.migfx.ch/ as a borker

Back to topics list

To add comments, please log in or register