I’m planning to check a method utilizing algorithms comparable to neural networks, described as follows
- Step 1: Learn the complete historic knowledge of 1 foreign money pair up to now for instance XAUUSD
- Step 2: Course of that knowledge right into a redefined knowledge, which is meant as enter knowledge for step 3.
- Step 3: Construct a logical algorithm that scans the information up to now and compares the information at the moment, then makes a shopping for and promoting determination
1. I’ll describe every step in additional element beneath
In step 1:
Re-reading knowledge from the previous is straightforward as a result of MT5 all the time gives knowledge from the previous for every tick
Nonetheless, this knowledge is giant as a result of it gives particulars in regards to the worth per tick, which in flip slows down the buying and selling course of.
So Step 2 will likely be wanted to course of this knowledge in order that it’s less complicated and lighter in measurement and quicker to course of
2. Course of worth historical past knowledge processing
In step 2;
First we’ve to outline what the final word goal of the information is.
On this case: My goal is to separate out which level to purchase, which level to promote, take revenue at which level, cease loss at which level, at the moment, what’s the RSI, MA, CCI, ATR Ask worth, Bid worth.
It is such as you watch a film once more and you may fully know the segments within the film from which you pick the required factors and save and create a extra concise film abstract (just like the Movie Assessment clips).
The that means of this remedy is: Assemble how conditions have occurred up to now, these conditions have clear solutions.
- The eventualities listed below are: RSI, MA, CCI, ATR, Ask/Bid worth
- The solutions are: Entry Purchase/Promote, Takeprofit/Stoploss
Decide the processing level.
For instance a highway 1 million meters lengthy, we can’t course of each millimeter. So let’s minimize it up each 1km and we’ll take a state of affairs there.
In my case: for each 1000Point will select a state of affairs
Create an attribute that classifies knowledge with a level of accuracy
How you can do: after creating the above knowledge array, we proceed to course of the information to be extra optimized as follows:
If the conditions happen and the outcomes happen extra typically and are comparable to one another to a higher extent, then the state of affairs is appreciated, i.e. excessive accuracy.
(Make your personal guidelines and laws for this evaluation.)
Right here for simplicity I solely classify with 3 ranges: Low, Medium, Excessive
This knowledge would be the mannequin to make use of for step 3
3. Course of knowledge and make buying and selling choices
In step 3:
Decide the processing level. Related: for each 1000Point will select a state of affairs
We are going to evaluate present conditions with previous conditions, if it’s the similar then make the identical choices because the outcomes had up to now.
Converse in additional element:
We evaluate the present indicators (RSI, MA, CCI, Ask/Bid worth) via all of the previous conditions created in step 2 i.e. RSI, MA, CCI, Ask/Bid worth).
If much like all indices with similarity, for instance higher than 90%, then execute Purchase/Promote, Takeprofit/Stoploss orders as up to now knowledge.
Be aware how dynamic you’ll be able to enable customization if you need.
On this step it’s going to occur 2 instances
The present end result is similar because the end result up to now knowledge, then we save the state of affairs and the end result once more into the previous knowledge array.
Present end result just isn’t appropriate, not like previous knowledge we appropriate this example and save end in previous knowledge array
Optimization: For quicker looking
Select to browse by class of historic knowledge accuracy first.
Evaluate with the previous state of affairs with excessive accuracy first, if there is no such thing as a case then go to medium stage, proceed to go to low stage, if no state of affairs add step 4
4. Refresh replace new knowledge
in step 4: We will deal with as follows, each 100,000Point ie experiencing 100 conditions, we are able to repeat step 1 and step 2.
The aim is to refresh the information, get new knowledge, and from there the information turns into increasingly more correct
It is the algorithm described in phrases:
I’ll depend on these primary descriptions to construct an automatic technique, through the development course of, there will definitely be many issues that should be dealt with, possibly the completion time will likely be longer than anticipated.
Trying ahead to your feedback and assist
Please comply with to obtain new notifications