expert testing of strategies
And that 4 figures and a test period is enough??? Show the system in action with the example of some expert...?
Artur, sorry, I couldn't find your email. I am writing it here. Here is a series by analogy with your example.
(+0,08
+90,04
-89,64
+0,32
+180,56
+90,12
+0,12
+90,08
+90,12
+0,12
+90,04
-2,08
+87,92
+89,48
+87,92
+89,48
-185,20
-94,68
-1,04
+360,00
+90,00
+90,12
+90,04
-89,88
+0,08
+180,08)
period = 8 weeks
Where is the author?
(-455.46
-464.1
700.02
4178.56
1961.29
-297.38
-306.56
-762.03
2031.36
-761.82
-237.18
-761.82
-411.23
-70.07
-507.88
1696.44
1172.3
-761.82
754.08
-762.02
506.62
-762.02
1017.1
-516.64
515.38
1331.33
314.63
672.34
79.14
-762.02
-499.12
86.9
-761.82
35.02
-656.41
3073.89
-499.13
1154.77
-761.82
1005.18
997.81
-332.41
3853.23
-761.82
-369.96
8047.96
-762.02
-796.51
718.23
-761.82
-578.78
-437.83
-674.43
-165.36
3721.12
3398.22
2151.9
8.75)
period = 10 weeks
Good day everyone, and thank you for your comments.
I will answer in order:
Figar0 - 4 figures is an example, I didn't want to give 100 figures for nothing.
KimIV, Parabellum - ok, I'll give you the results tonight. I can not respond instantly online, every response takes time, I do not keep the software on my home computer.
My address is fxforum dog inbox dot.ru (I have not found how to make an email visible in the profile, advise how)
Until contact.
KimIV: score 2, increase the number of trades to 100.
Parabellum: score 3, increase the number of trades to 100.
You guys have great scores, I advise you to do 20 manual trades at least in offline mode (i.e. if at the time of entering the trade we were sleeping, then count as if we entered - paper trading is called), but not on history and using quotes feed (free) from some western supplier. For example choose from those offered here: dailyfx dot com chart section. It will take you a month or two, and if the result repeats, as they say, welcome to the Caribbean...
And in general, paper trading is the toughest instance. Sometimes in paper trading it turns out that what we put in the program is not realized in practice due to ambiguities of interpretation, which due to the nature of program writing are interpreted for better and the profit is actually obtained through inaccurate or erroneous programming - these nuances should be kept in mind.
The quotes history is also important. Imagine that you are running your simulations based on quotes you have read out of thin air. what's the price of that work? .... did you try to check the validity of quotes used ... Actually that is what paper trading of imported flows is for (no guarantee either, but the more variations, the more stable the result).
M-nya.... Well, why make manual trades when you can put an expert on the account? Or run a forward test...
I do not understand what is the purpose of this attachment? To put a "black box" in a "black bag" for a fee and for what? Why should the results of your, I don't even know what and there is no name for it, be trusted more than the test results? It may make some sense, but not for automated strategies. IMHO.
"I'm not evil, I want to understand."
"I'm not evil, I want to understand."
And in general, it's strange.
Even purely visually, the respected KimIV has a more stable result and a lower score...
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
Good afternoon, everyone.
I would like to offer advanced EA writers a tool for evaluating strategy stability. As many people probably noticed, most strategies that have demonstrated excellent results on history, either have poor performance in real trading, or lead to losses. As it turns out, neither the percentage of profitable trades nor their amount, let alone their absolute value over the testing period, can guarantee their stability. As we have seen, even strategies with an 80% profit can make a losing account. And then the question arises as to whether there is a mechanism to evaluate a strategy based on its historical performance.
First of all, this problem is of interest for large banks and hedge funds, which practice trading with so-called black and grey boxes. By the way, trading with trading systems (black and gray boxes) came precisely from this environment.
A unique program that allows you to determine the probability of its stability in the future, based on the results of the expert's work on history, has become available to us. The beauty of the assessment lies in the fact that no knowledge of the principles of operation of the Expert Advisor, much less of the code, is required.
The expert evaluation is given on a five-point scale:
1 - the strategy is likely to fit the curve and does not show sustainability;
2 - the strategy shows signs of sustainability and should be paid closer attention to during further testing;
3 - the strategy shows stability and should be paid closer attention to during further testing;
4 - strategy shows stability, you can work with it on real money;
5 - the strategy has excellent performance, you can work with it for real money.
It should be noted that even a score of 2 is encouraging on this scale.
All that is required for testing is a table of results indicating the time period over which the simulation was run. An example of input data is as follows
(+10
-15
+50
-30)
period = 1 week.
In brackets we see the usual vector of results expressed in pips or already in a specific currency.
At the initial stage it is proposed to use the described evaluation "for free". Those who are interested, please send an email.