From theory to practice - page 1148

[Evgeniy Kvasov]

secret:

Yes, you need 10-20 times as much.

Typical return system equity is several small profits and one big loss. To estimate the system behavior in case of losses, we need to have at least 20-30 losses in the equity in order to estimate the minimum statistical reliability.

If the author knew how to use the tester, he would not have spent years of his life on checking in the real account.

Well, sometimes it is difficult to modify what has already been written for verification in the tester. Sometimes it is difficult, sometimes lazy... Sometimes you can see how it works) I decided to test it this way... OK, it's just that there's an interest group, the worthy ones are promised .........)) And besides, he calls me felix all the time))))))))))

[Anatolii Zainchkovskii]

vladevgeniy:

Well, sometimes it's hard to rework what you've already written to check in the tester. Sometimes it's hard, sometimes it's lazy... Sometimes you can see how it works) I decided to test it this way... OK, it's just that there's an interest group, the worthy ones are promised .........)) And he calls me felix all the time))))))))))

he must have felix the cat and you have a kitty ava)

[CHINGIZ MUSTAFAEV]

If you don't know the mechanics of price movement, it is absolutely useless to do anything at all. If the nuances of estimating the mechanics of price movement are reflected in the algorithm, then everything is fine - the result when the parameters change has a linear dependence, otherwise there will be chaos and nothing else.

[Alexander_K]

I would like to say one thing, lest anyone think that everything is already clear and straightforward and you can safely line your pockets with cash.

Yes, the results so far are very good, but:

1. not enough statistical data, you need at least 100 trades made. I`m afraid I`ll have to sit on a demo for May as well.

2. I cannot speed up the process and "test" the TS in a tester - I have a non-standard way of receiving a sparse stream of quotes and the archives are simply nil.

3. Yes, I use nonparametric methods of variance, kurtosis, skewness and weighted average with some weights. These are all interesting, etc. But, to the conceptual questions, viz:

a) why is the distribution of increments in the market exactly like this and not, for example, Gaussian?

b) Why, in my case, does the price, after all, return to the average in most cases?

I can't answer that.

I take it as a given, confirmed experimentally. I cannot theoretically justify the answers to these questions.

Is there a possibility of an embarrassing plum in this case? Well, no, I don't, but there is a chance of losing 1 or 2 shameful deals per month.

I'm not making excuses in advance - just the above questions need to be answered.

For now - we just watch.

[Renat Akhtyamov]

Alexander_K:

Is there any chance of an embarrassing loss in this case? Well, no, but there is a chance of losing 1-2 deals a month.

there's

and every single one of them

in the picture.

https://www.mql5.com/ru/forum/297446/page173#comment_11310708

[CHINGIZ MUSTAFAEV]

Alexander_K:

I would like to say one thing, lest anyone think that everything is already clear and straightforward and you can safely line your pockets with cash.

Yes, the results so far are very good, but:

1. not enough statistical data, you need at least 100 trades made. I`m afraid I`ll have to sit on a demo for May as well.

2. I cannot speed up the process and "test" the TS in a tester - I have a non-standard way of receiving a sparse stream of quotes and the archives are simply nil.

3. Yes, I use nonparametric methods of variance, kurtosis, skewness and weighted average with some weights. These are all interesting, etc. But, to the conceptual questions, viz:

a) why is the distribution of increments in the market exactly like this and not, for example, Gaussian?

b) Why, in my case, does the price, after all, return to the average in most cases?

I can't answer that.

I take it as a given, confirmed experimentally. I cannot theoretically justify the answers to these questions.

Is there a possibility of an embarrassing plum in this case? Well, no, I don't, but there is a chance of losing 1 or 2 shameful deals per month.

I'm not making excuses in advance - just the above questions need to be answered.

For now - we just watch.

One picture and the answers would be in your pocket - funny isn't it:)))

[secret]

Alexander_K:

2. there is no way to speed up the process and "run" the TS in the tester - I have a non-standard way to receive a sparse stream of quotes and such archives are simply not available anywhere.

The second bars are easy to make from ticks yourself - google FXT files.

[secret]

Alexander_K:

So that no one thinks that everything is already clear and understandable and you can safely line your pockets with cash.

With cash another surprise awaits you - according to Russian law, the Forex tax must be paid on the amount of profitable transactions, without taking into account unprofitable ones.)

Of course, you can pretend that you do not know this and pay the total, but it's a matter of luck, until the first check everything will be fine).

[aleger]

secret:

With cash, another surprise awaits you - according to Russian law, forex tax must be paid on the amount of profitable transactions, without taking into account unprofitable ones).

Of course, you can pretend that you do not know this and pay on the total, but it's a matter of luck, until the first inspection everything will be fine).

This law is at least 15 years old.

[Evgeniy Kvasov]

secret:

With cash, another surprise awaits you - according to Russian law, forex tax must be paid on the amount of profitable transactions, without taking into account unprofitable ones).

Of course, you can pretend that you do not know this and pay the total, but there you are lucky, until the first check everything will be fine).

(aksumoron - russian laws)))