Poll: balance of the winner - page 7

New comment
 
You guessed it :)

By the way, your first phrase the MT authors and all the brokers can also take it personally ... :(

But I believe that someone will get prizes.
And the system will earn this award,
But in the next 3 months, it'll probably lose that profit,
but it doesn't matter...

I'm not bullshitting anybody.
CurveFitting is the problem with all optimized models.
You can fight it, but it's hard and with varying success.

All participate here for good luck, and it's not a freebie, but an advertising campaign.
And the more participants you have and the more good systems you have, the more successful it will be.

SO WHAT ARE YOU OFFENDED BY ...

I do not advertise my GO, I have not given the name of the package or links.
And the reward is not counted on, if my system will be in the black, then a small one.
The results that are declared here it is not light.

Moreover, I did not have time to debug it properly, and now I see that all sorts of glitches are creeping in.
 
Mak писал (а):

WHAT ARE YOU OFFENDED BY ...?

Don't you get it? - On the WISE phrase about other traders fitting their Expert Advisors on historical data. When it is said by beginners, you can understand it, but when it is said by long-time traders, you can only be offended. You do not allow people to dream. If it were a debriefing during the Championship or at the end of it, the work on the mistakes would be useful for everyone. If you are convinced that the Experts will not reach the finish line on historical data, then why talk about it in advance? Or are you afraid that your untuned Expert won't reach the finish line either and don't want to be on the same step with the newcomers? At least be smarter than the beginners, even with clever expressions.

Let's have a look at the Championship results first, and then analyze how the winning Expert Advisor reached the finish line. In this thread, people are only commenting on the possible results of the winners and NOT MORE.

 
Well, I was talking about the possible results of the competition.
And for some reason you decided to lecture me about optimisation.
And speaking of my CS, I just wanted to show you that I have long been on the subject.

If you do not understand what others say, it does not mean that they are clever.
You just don't get it ....

I wasn't referring to other expert writers talking about fitting.
I was referring to the peculiarities of the optimization process of any models.
The tendency of that process to CurveFitting, and the fact that the results
after optimization usually have nothing to do with the actual results.

By the way, the debriefing can already be started.
The contest has already started, and the participants cannot change anything.
 
Mak писал (а):

By the way, the debriefing can already begin.
The competition has already started and the contestants can't change anything.
Good. I'm done talking about it. Let's know you're on board.
Good luck with that.
 
I am currently writing a theory. I can tell you one thing: if an Expert Advisor is "fitted" on historical data and makes a large number of trades and has an even profit, then it has a better chance of winning, but much depends on the fitting period.
The proof is simple: let us assume that we have adjusted on a half year interval. It can be broken down into the first 3 months and the second 3 months. It turns out that the same strategy was in effect for 6 months. Now think. What if some Expert Advisor has found this strategy in the last 3 months, before the championship?) The basic conditions have to be fulfilled (see above:)) That said, I will note that there are many such strategies.
 
The theory will work if there is a smooth transition of parameters. Unfortunately, there is no such thing. Forex is constantly proving its randomness. The only thing that has been proven is that there is a small trend component on large frames.
 
I disagree. Much depends on the methodology of the expert. Although forex is random, random numbers also belong to distribution laws.
 
Come on, philosophise. Relax and enjoy yourself. I'm not going to worry so much anyway, because I know what I have to do in the next six months. After all, the championship is not the goal. It has given an impetus to a lot of developers to work intensively. And the results of intensive work are often outstanding.
As for the results of off-line and on-line testing, I can say that the differences are obvious. When I look at my program, I see that there is still a difference between the opening and closing price in the tester and the demo. It is not considerable, one or two points, but it may occur. The program copes with it. I have not noticed any difference in time of opening. But if any program is critical to one or two points, it will certainly fail. I hope mine won't crawl under the plinth and embarrass my grey head :) I don't really need anything else from the championship in principle. Just a quality check. Anyway, the championship version is different from my battle one. It is more truncated and simple. I will correct the basic version as well, taking into account my experience in the championship.
 
Mak:

If you don't understand what others are saying, it doesn't mean they are being clever.
You just don't understand ....


Everything you write is clear. My entry is not optimised on history and I'm not looking for the best parameters for it in the tester. But I need to decide on trailing stop and stop loss. The tester helps me well here. I think it will help me to determine the best parameters on 1-minute chart. Will there be CurveFitting in this case? I have provided the test data, but it is clear that it is curved because of the entries. The inputs will be different and so will the results, of course. Now you can tell me why.
 
Mak:
IMHO, fully automatic trading systems can exist,
but they will be big monsters that MT cannot cope with ...
The ideology will be different there too ...

Good where we are not :) If you connect the Hydrometeocenter to all the supercomputers and hire the best programmers, the accuracy of weather forecasts will not improve. API or computing power has nothing to do with it. For any complex system the accuracy of forecasts decreases with time, sooner or later reaching a state where even opposite events are equally likely - this is the limit. The only question is whether it is possible to work out price movements up to that limit, getting more than the size of the spread on arbitrage. If you can, you can use a calculator to do the math.

And what is easier and also makes sense are so-called DSSS (Decision Support Systems).
(Decision Support Systems).

If everything goes wrong, you can blame everything on the decision maker, while the Expert Advisor has nothing to do with it :)
1234567891011
New comment