For those who are convinced that all EAs with a martin are losing out. - page 42

 
Mathemat:

Don't confuse trading on minutes and tests on minutes. I specifically stressed that


I am not confusing anything.

Again. If you trade on, say, an M15 open and test on M1 you will get not tests, but a sham.

 
GEFEL:


It is clear. But don't you have enough margin for mini-profits when trading the candlestick shadows? Such a reserve must be available.

And if you trade on oupens and test on ticks, they are two completely different processes. Neither the number of trades nor their results will be the same as the actual trades. Then if you test by ticks, you will measure some abstract drawdown. Do you need it?

I'm testing on oupens. I did it by ticks once to see how different the maximal drawdown is. Drawdown by ticks is not abstract. In real trading it is based on ticks. The important thing is not the ticks themselves, but that price passes both low and high with ticks, and not only open. This affects the drawdown value. The results will never coincide with the test ones because quotes during testing and on the real account are different.

 
GEFEL:


.....

As far as minute tests and even more so trading on minutes - I think this is a widespread misconception. In this case you are trading noise and nothing but noise. And this is the destiny of kamikaze.



The ticks are really coming to your terminal. Think about it.

So you don't have to tear your hair out over the reasons why the test was OK and the actual tics went to hell.

 
khorosh:
It would be nice if the tester could test not only on the open, but also on the high and low. Then the maximum drawdown results would be the same as when testing by ticks.
Test by control points.
 

If testing is to be done as close to a real trade as possible, then it should be done:

- Test on a real tick history collected or provided by a method of 99% simulation quality.

- Make all trades only on ticks that were in the real tick history, alternatively - on the next tick ( 1 - 5 ticks ), this applies to any transactions and stop loss and all other types of orders, only then will take into account all slippages on gaps (no ticks - no transactions) and spread widening.

In alpari for example, recently there appeared an opportunity to download ticks for the whole period
http://forum.alpari.ru/showthread.php?t=79168
http://ticks.alpari.org/

 
khorosh:

I test on the oupen. And I ran it on ticks once to find out how different the maximum drawdown is. The drawdown on ticks is not distracted. In real trade the work is based on ticks. The important thing is not the ticks themselves, but that price passes both low and high with ticks, and not only open. This affects the drawdown value. The results will never coincide with the test, because quotes during testing and on the real account are different.


Things are even worse than they seem. It turns out that the Expert Advisor is working on ticks while it is tested on oupen M15. So those pictures and reports at the beginning of the thread are worthless?

 
paukas:

Realistically there are ticks going to your terminal. Think about it.

So you don't have to tear your hair out to find out why the test was fine and the real one went to hell.


So let them go. And I filter them out and only trade on the oupen. My tester results pips to pips are the same as

Real, and yours never do. Take care of your hair.

 
GEFEL:


So let them go. And I filter them, and only trade on the oupen. My tester results pips to pips are the same as

by the real world, but yours never will. Take care of your hair.


That's right, only in reverse. Mine do. Your results don't yet.
 
paukas:

Right, only in reverse. Mine are the same. Your results are not there yet.

If you mean the number of posts, you win. -;)
 
GEFEL:

If you mean the number of posts, you win. -;)

It's not a competition, it's a forum, sweetheart.

I've warned you about testing, if you want to step on the same old rake, I won't force you. Off you go.

Reason: