"The 'perfect' trading system - page 146

[MIRAAL]

Angela:

Everyone knows that nothing in the world is perfect, but in solving a problem we strive for the ideal. But to get closer to it, we should describe it at least verbally and formalize it in the form of goals and ways to achieve them, otherwise we end up going there - without knowing where, looking for something - without knowing what. If we create TS and want to improve it, we need to define the ideal to which we strive, to make rules according to which an "ideal" TS will work (hereinafter referred to as ITS for short). Of course, we can say that ITS should bring maximum profit and minimum losses. But in this thread I put the question somewhat in a different plane, what rules should follow the ITS in its work to achieve this. In this topic I propose to make a set of rules for ITS to work, and formalize these rules in the form of algorithms for different classes of TS (under the classes in this case should be understood different principles underlying the TS, for example, assay - channel, fractal, etc.). If anyone is interested in such a formulation of the task, join the discussion, and with joint efforts we will be able to detect more nuances and crystallize them in the form of concrete principles of operation, common for all classes of TS, as well as take into account specific features of separate classes. And by developing a "code of honour" for ITS, everyone will be able to take these rules into account when designing their own TS in accordance with their tasks.

In order not to be unfounded, I will begin by putting the formulated rules in bold type so that they can be easily seen in the text.

1). One ofthe most important principles underlying ITS is its ability to learn and adapt to the changing phases of the market. And secondly, 2). The minimum number of parameters regulated from outside. To determine the principles of self-learning ITS we should probably talk about a particular class of TS. For a start I propose to consider a class of TS working on a breakdown of a channel. The question arises, how do we build the channel? We can use various indicators for that, but then a new task will appear, to make ITS adaptive to the market we should autotune indicators, decide on the criteria defining their state in the market and parameters that should be regulated and, most importantly, by what laws to adjust the parameters. In order to solve this problem, we should be more specific about the type of indicators to be used, and it will be our own challenging task. The option of self-learning of ITS without the use of external indicators is also possible.

For example, I propose to solve the problem of self-adaptation using learning of ITS on virtually executed trades on the history interval, located directly at the zero bar. For this purpose we need to set two external parameters:

1. The necessary and sufficient depth of history for learning (though this parameter can also be set in the selftuning mode later);

2. The minimum size of trades that we allow to trade. The shorter is the target for the TS, the more profit it is able to obtain, but there is a limitation, which we should consider and set in the external settings, not every brokerage company will tolerate a long trade with targets under 10 points.

To be more specific, I will use the picture below as an example:

Fig.1

The trading channel width can be calculated as an average value of N ideal trades built, for example, using Adept block given by me in another topic (working results in Fig.1) or any of Zig-Zag variants, though it will be less flexible. Next, determine the timing of opening virtual positions, e.g. when the 30% level is broken through from the midline of the channel, and the timing of closing them, there may be several options here:

a). if the border of the channel is reached;

b). if the level below the middle line of the channel has dropped below the line;

c). upon breaching the channel boundary and exceeding the level by 1/2 of the channel width, transfer the level of the average to the channel boundary, and control the position relative to the new average.

There may be many other options - suggest them.

When making Q - virtual trades, we analyze them in order to develop the rules of self-training of the TS for specific market conditions. To be continued ....

And meanwhile, offer your variants of rules of ITS, and make additions and corrections in the material stated by me.

You are going in the right direction! Since the very beginning of my acquaintance with mql4 I have been using the principles of self-adaptation of Expert Advisors. It's true I've encountered a couple of pitfalls)))

[Vladimir Paukas]

VictorArt:

Too bad you had the chance to drain your PAMM, but denk didn't let you take it.

I wonder if you would then talk about yourself as frankly, about shoehorned investors? :)....

Do not wishful thinking. There was no chance.