Avalanche - page 427

 
lasso:

Easy! And what is it? ))


Well about 45 degrees from $100 to ... )))

The dynamic channel can be added to https://www.mql5.com/ru/code in a few minutes - to test this strategy this code is good - simple and clear, do not forget to change the starting price at the closing of all orders to the current one.

SZZY: it is interesting to look at real trading - in one brokerage company not bad monitoring (screenshots are like in the tester) right next to the home page - that's where it is really interesting, everyone is different, not that Slaw....

 
IgorM:

The dynamic channel can be added to https://www.mql5.com/ru/code in a few minutes - to test such a strategy this code is decent - simple and straightforward, do not forget to change the starting price when closing all orders to the current one.

SZZY: it's interesting to see the real trade - in one brokerage company monitoring is not bad (screenshots are like in the tester) right next to the home page - it's really interesting, everyone is different, not like Slaw....

But recently I've given up writing EAs as functionality for testing of trading strategies. In other words: there is a price (series) and there is an entry point and direction (and even that is not always the case), the first task is to find a reliable statistical advantage (positive expectation, inefficiency) using statistical methods and only then we can write an EA.

And it will not have to be optimized. Its task is to trade, to execute logic.

And the dynamic channel is determined by some parameters, it means that also the adjustment is not excluded.

I have been hearing some talks here about EAs without optimisable parameters, now I think I understand what these people have been writing about.

P.S. Thank you for the link. It will come in handy...

 
lasso:

A dynamic channel is defined by some parameters, which means that we cannot rule out fitting.

I have sometimes heard here about EAs without optimisable parameters, now I think I am beginning to understand what these people were writing about.


Tell me more or a link or ... - I'm interested in it, I've read a lot of old forum threads, maybe I've missed something, please send a message.

ZS: dynamic channel should definitely be with parameters, because the volatility of each currency at different season, time is different (I pointed out to you that I tried to determine on M5), because I think that such simple strategies like "Avalanche" - trade the current value, in fact - bet on whether this volatility (channel width) or that - but this is my vision of avalanche-like EAs

 
lasso:

But recently I've given up writing EAs as functionality for testing of trading strategies. In other words: there is a price (series) and there is an entry point and direction (and even that is not always the case), the first task is to find a reliable statistical advantage (positive expectation, inefficiency) using statistical methods and only then we can write an EA.

And it will not have to be optimized. Its task is to trade, to execute logic.

A dynamic channel is defined by some parameters, which means that we cannot rule out fitting.

I have heard some suggestions here about EAs without optimisable parameters, now I think I understand what these people were writing about.

P.S. Thank you for the link. It will come in handy...

I couldn't resist... I agree, bull's-eye.
 
IgorM:

more details or a link or ... - Very interesting to look at, I'm interested myself, I've read a lot of old threads on the forum, maybe I missed something, you can send me a message.

Statements on this implementation have always been casually in the threads about optimization, and I can't recall a substantive discussion or a separate thread, unfortunately. Maybe the coryphaei can advise?

............

If you want more details about what I have - everything is simple. ))

We take a signal (any kind, from simple to sophisticated), set a grid of levels (I have five levels in each direction multiplied by three typical strategies) and use it to track the achievement of these levels independently from other signals and for each strategy. As a result of the script processing we obtain statistics on the "quality" of the signal.

If it turns out that the signal is random, then what is the point of waiting for it? Just a waste of time?

For example I've been working with stochastics, but I've got nothing. Although, at first I liked it, and the forum members approved this indicator ....

And Avalanche.......????????????????????????????????????????? Madam is very interesting. ))

 
lasso:

There have always been statements about this implementation in the optimization threads, and unfortunately, I don't recall a solid discussion or a separate thread. Maybe the coryphaei can suggest it?

............

If you want details of what I have, everything is simple. ))

I take a signal (any kind of signal, from simple to sophisticated), set a grid of levels (I have five levels in each direction multiplied by three typical strategies) and use it to track the achievement of these levels independently of other signals and for each strategy. As a result of the script processing we obtain statistics on the "quality" of the signal.

If it turns out that the signal is random, then what is the point of waiting for it? Just a waste of time?

As an example, I recently "skimmed" stochastics but did not find anything. Although, at first I was attracted to it, and forum users approved this indicator....

And Avalanche.......????????????????????????????????????????? Madam is very interesting. ))



I am also interested in Avalanche, but have to watch it all the time.
Mostly works at night if there's no news.
I cannot think of parameter(channel break or some indicator) that would disable it in case of high volatility.
Maybe you have some ideas?
 
Stells:

I'm interested in Avalanche too, but have to follow it all the time.
It mostly works at night when there is no news.
I can't think of a parameter (channel break or any indicator) that would disable it in case of high volatility.
Maybe you have some ideas?


Maybe we are using different algorithms. In my opinion, on the contrary, in low volatility the avalanche corridor should be maximal, while in high volatility it should be minimal.

The question is how much time and where. ideally we wait for the signal and in 1 - 2 flips we will take the jackpot.

martin flip - in general, it's an accumulation of lots to work in a powerful trend.

When the Rx is close to 100 and zero, the avalanche corridor is minimal since it signals a strong move. At Rx 50 it is minimal since we are sure that we are in a tough flat.

 
This is an indicator for building adaptive channels, I made it as a part of Borispolz's topic. It depends on the market and is built on the basis of regression. There are two parameters there, first is the maximum number of bars on which the channel is built, the optimal channel is not necessarily built on this value. The second parameter is the maximum deviation from the regression line for finding support and resistance lines. Unfortunately, calculations are a bit heavy. If you know how to do it, drop me a line or two.
Files:
 
I screwed iReg better, but the loss is inevitable. I tried to expand the channel depending on the number of orders, the same thing. I have stopped at hedge - at minus 100 the EA goes into hedge and waits for the master to destroy it. I need a good movement or trend indicator.
 
If such an indicator existed - - then why bother with the other systems - just wait for the signal and put the entire deposit in the forefront... :) a billion dollars a year - guaranteed :)