This trading Expert Advisor learns from its own mistakes. Each unsuccessful trade is thoroughly analyzed and turned into a future success. The trading EA uses the time efficiently working 24/7. When the market is closed, the trading EA is especially productive, as it analyzes its own mistakes and makes a better trading plan for the coming week. All this thanks to its self-learning intelligent system.
Full version: https://www.mql5.com/en/market/product/967
Free version: https://www.mql5.com/en/market/product/4035
In the full version:
- There is an integrated (trained, experienced) memory's for many symbols (financial instruments). The list can be found in the version history.
- Maximum power.
- The maximum concentration level when learning: unlimited.
- There are abstract money management. Can trade with large volume.
- Possibility to regulate the level of experience, from which the EA allowed to actively trade.
- Advanced settings, extra settings.
Price Tension Trading Strategy
The trading Expert Advisor learns by using its trading strategy. The strategy here is a tool simplifying analysis and decision-making. During the learning process, the EA can modify the strategy by himself to a certain degree for greater flexibility. In future updates will be modified and enhanced trading strategy.
The heart of the trading system is an algorithm that identifies the favorable moments for trading. These points are formed by the price tension relative to the normal state that allows making timely trading decisions.
Each deal is protected by stop orders (Stop Loss and Take Profit) with a minimum distance of 150 points (for the quotes with 5 or 3 digits after the decimal point) from a position open price. Position opening side may be changed without the need to wait for stop level activation.
Such methods as averaging, martingale and as well as scalping are not used. Mid-term trading is applied instead. However, sometimes, the EA can change its trading style - the self-learning system is able to adapt its style to market conditions.
The trading EA performs 5 trades per week on the average. The mathematical expectation is 150 points (for the quotes with 5 or 3 digits after the decimal point), so indicating its weak dependence on the spread and ping to the trading server.
Self-Learning Intelligent System
The system works like an autopilot with the only difference that it can develop and learn in new situations.
- The main objective is to reach the optimum reliability by improving the efficiency of trading decisions.
- The second one is to minimize losses. Security is an important element for the EA. Therefore, mid-term trading plays a major role in its activity.
- The third one is to make the income more reliable by increasing the average profit and decreasing the average loss.
The system learns both from practical mistakes and from theoretical mistakes. For greater security, it tries not to make practical mistakes while learning, trying to learn on a theoretical mistake, for this it to theoretically simulate a variety of situations and is looking for solutions for them, in practice, compares the theoretical knowledge to improve the simulation and improve their realism. The EA uses both trading results and its own virtual space where it can check its hypotheses and put them into practice in more secure manner. Virtual learning is one of its main advantages, as the trading EA can analyze a deep history with a plenty of market cases.
The Expert Advisor requires at least 7 years of historical data (minute bars), as it analyzes a quite large number of bars to make a decision. The larger the history and the higher the quality, the better. During the launch, the trading EA will try to self-check your historical data of minute bars. However, if the broker does not possess the necessary history depth or if the quality is too bad, the EA will not be able to learn and make decisions properly.
To run in the strategy tester there is an important nuance, which is described below (in the "Strategy Tester"), do not miss it, please!
Experience is the product's main development parameter expressed in units. The more time the EA spends on trading and learning, the more experienced it becomes. Another important parameter is concentration: the higher and more reliable this parameter and the higher its level, the deeper its concentration making learning more fast and efficient.
The system learns 24 hours a day, 7 days a week. Operations speed has been optimized well enough in order not to load CPU to the full.
I recommend that you do not disable the Expert Advisor on weekends, as this is the most productive period for the EA when it can improve its trading plan for the coming week. Otherwise, skipping the weekends the system will be to develop a trading plan in stress mode!
Question-answer
How optimize in the strategy tester?
Do not need. It’s system independently, flexibly and adaptive compared with linear.
Have the EA a trading strategy?
Yes. Called: "Price Tension", so EA learns to manage it.
EA can learn, but not trade, or both learn and trade?
Yes. EA learn in all the time, whether it trades now or not. Only in extreme cases it can enter in the cryostasis: if there is a failure or a minute bars history of the terminal is not enough of quality or if other exceptional situations.
Then how the EA will be able to learn, if not to be trades?
The fact that the EA is able to learn using historical datas of minute bars (note that it is necessary historical datas, at least for the last 7 years. And the more history available, the better). For it, this so-called learning in the virtual space, where it comes up with hypotheses and there make checks them, due to this, it manages to come up, test and learn more hypotheses than if it only relied on the trade results.
But if the EA learns all the time, and thus develops, so how it cannot forget the development?
It has a memory. Memory is stored in a special file on your hard disk, and updates. Even if happen an emergency stoping of the EA, the memory will not disappear and when you next start the EA, it will remembers and continue to learns and trade.
What Time Frame?
It is desirable a minute time frame (M1). For the Strategy Tester necessarily a minute time frame (M1). For online come down any.
Memory.
All experience, hypothesis, ideas and goals it stored in memory, which implemented in the form of a file and stored on your hard drive disk. Even if happens an emergency stops of the EA, the memory will not disappear and when you next launching the trading EA, it will remember and continue to learns and trades, though it will have to spend some time to reach the same depth and concentration of the level.
Memory File is stored in a shared folder of all MetaTrader 5 client terminals, usually (for Windows 7):
c:\Users\<user name>\AppData\Roaming\MetaQuotes\Terminal\Common\Files\
Next in the subdirectory:
\Price Tension\MEMORY\Accaunt <account number>\
<account number> - the number of your account.
File memory:
Memory [Price Tension Free] [<symbol>].ptf
<symbol> - name of the symbol (financial instrument).
Do not rename or change the memory file, otherwise it will cease to be correctly and usable.
For each symbol (financial instrument) is intended its own memory, because every instrument develops separately by the EA. By the way, the EAunderstands symbols with prefix characters.
If you want to continue to develop the old memory on an another account (for example, if you changed the account number), you must manually move the file to the folder with the account number (you have to create it if it is missing).
The EA have the integrated (already trained, with experience) memory's. If the file is damaged or absent, then the EA first will try to load the backup of the memory file (if this file previously created by the EA), otherwise it will attempt to load the integrated memory, If the integrated memory for symbol (financial instrument) in the EA is absent, then it try to load a clean memory (empty memory, without experience). Integrated memory is different from a clean memory: the integrated memory is prepared (trained) already and contains some experience, sufficient for practical work.
Full list of integrated memory's, you can find in the history changes ("What's new" tab) of the product.
Strategy Tester.
You can run in the strategy tester, but it is necessary to consider some of the nuances.
First, you can't use the optimizer. This is not necessary.
Second,
For a complete decision-making requires at least 7 years of historical data (minute bars history). But when EA start runing in the strategy tester, initially available insufficient depth of the historical data of minute bars. For the accumulation enough of historical data to analysis, the EA needs have to run from an earlier date relative to the input setting (in the input settings of the EA): "Starting date of the activity".
So if you want to get the result, for example: for 2016 year (2016.01.01 - 2016.01.01), you should the setting "start date" of the Strategy Tester to set less than or equal 2009.01.01, and the setting "Starting date of the activity" (in the settings of the EA) to set = 2016.01.01.
Third, there is another nuance of working in the Strategy Tester. The EA can take some times to analyze, especially if needs to deep make a trading plan or if you specified «Max. concentration level» > 0, so for the one trade week you'll have to wait from tens seconds to several hours. Of course, in a online working it take a lot of more time, since in the online working it tries to find the best concentration, that gives more accurate trading decisions.
By the way, it is not have to run on every tick, because the EA only analyzes the formed minute bars.
Online working and trading
The EA understands all account types. Adapted to trading on real accounts.
There are additional features such as: spread filter, several attempts of trade requests when broker reject a trade request (for example: requotes), sending messages to e-mail, push-notifications etc.
If you are using a hedge account, then you can use the magic number, so you can isolate the trading activity of the EA from other EA on the same symbol (financial instrument).
But if you decide to use a netting accounts, then I strictly recommend you to not launch a several experts on the same financial instrument at the same time, otherwise there may be conflict when working with the trade position. By the way, the EA understands symbols with prefix characters.
EA have the intellectual system of the analysis environment, therefore if market is closed or there are restrictions on trade, then the EA can identified and considered these situations. At startup, the system will try to perform historical data (minute bars history) in your terminal. If they are insufficient or the quality is too bad, then the system will not be able to continue their activities.
For a complete decision-making requires at least 7 years of historical data (minute bars history). So you need to make sure, that your broker has enough of minute bars historical data and the quality also is enough.
Other
- My thoughts, plans and ideas for the development of this project: https://www.mql5.com/ru/blogs/post/463938
- Description of the update (v.5.123): https://www.mql5.com/ru/blogs/post/688859
