"New Neural" is an Open Source neural network engine project for the MetaTrader 5 platform. - page 13
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Avals:
How to train the TS we need, if it should not work without the Boolean part (giving profits or predicting price changes)
Let's use "you" and an example. For example, let's build a hypothetical (but concrete) nontrivial TS and try to solve it on neurons. Also hypothetical.
You may try to use that TS you posted at forum 4. For example, to enter on the breakout, and the horizon levels to determine automatically.
Let's start with "you" and an example. For example, let's build a hypothetical (but concrete) nontrivial TS and try to solve it using neurons. Also hypothetical.
ok. The NS should be a trend filter and direction of movement. Entry at the breakout of the local extremum, then transfer to Breakeven. TP, SL, the order of extremum and transfer to Breakeven are parameters. How to train the NS?
ok. NA should be a trend filter and direction of movement
So in order. We reject the trend filter for the time being, first of all, we should define the concept of trend/non-trend, and secondly, it can be done separately.
We remove the transfer to Breakeven, it is an independent part.
Optimal SL TPs may be searched at the same time as the entry horizons.
So in order. We reject the trend filter for now, first of all we need to define the concept of trend/non-trend, and secondly it can be done separately.
We remove the transfer to the break-even, it is an independent part.
The trend filter is the NS.
Simply, if the MA gives +1, we trade only a break-up upwards according to this scheme, if -1, we trade only downwards. 0 - we do not trade.
Breakeven is not a separate part and it is dependent. Again, from the context, as Sweeney likes to say, and the context in this case defines the NS
P.S. But for simplicity I agree to discard :)
In simple terms, if the NS gives +1 trade only a break up this scheme, if -1 only down. 0 - we do not trade.
Hm. Why not trade if the horizons will give plus?
Breakeven is not a separate part and it is dependent. Again from the context as Sweeney likes to say, and the context in this case defines the NS
Well, hello. Where in the original TS is the definition of context? And if it is dependent, then the rules of the set of BUs in the studio.
Neuronics itself doesn't owe anyone anything. What we set is what it will produce.
Ahem. Why not trade if the horizons you give out will give you a plus?
I do not understand about the horizons, but we can simplify: NS gives us a recommendation that the system should be traded now only long or only short.
Hello. Where in the original TS is the definition of the context? And if it is dependent, then the rules of the positioning of closed positions in the studio.
Neuronka itself does not owe anyone anything. What we set, it will produce.
The NS defines the context for the system - when to trade and in what direction. The BU algorithm may be of any kind, but for the sake of simplicity it won't exist (transfer into BU).
The neuronics should identify the periods of trade that are favorable to the system on the basis of price series
The NS determines the context for the system - when to trade and in which direction.
ok. The NS should be a trend filter and direction of movement. Entry at the breakout of the local extremum, then transfer to Breakeven. TP, SL, the order of extremum and transfer to Breakeven are parameters. How to train the NS?
Simply, if the NS shows +1, we trade only a break-up upwards according to this scheme, if -1, we trade only downwards. 0 - we do not trade.
1) There is one bug in the scheme of exits [-1;0;1], in fact all three ways out should be equal and it is very difficult to hold a hypertangent on zero or a sigmoid on 0.5 and it will want to jump.
2) Bulashev's"Statistics for traders" has a scheme of position (order) efficiency evaluation. You can apply this scheme and train the net to provide trading signals while trills and breakeven are all elements of the TS not related to the net.
3) Filters are elements of preprocessing (preparation of examples). If you put preprocessing into the grid algorithm, universalization will not be attainable.
Okay, how is the context set?
I want to teach NS to trade by your TS, adding a couple of degrees of freedom to it.