mql5/Experts/Advisors/DualEA/PaperEA/### Automated trading algorithms.txt

### Automated trading algorithms

Using expert advisors is a simple task today. The fantastic development of MetaTrader 4 and 5 allows us to spend no more than a couple of minutes to download, test, and run an automatic software to trade our money. Thousands of expert advisors made by hundreds of authors are available online, and to buy one, you need no more than two clicks. A positive backtest report makes you feel that you met the right software. Then, you install it on a demo account and run it for a while. If it is working with virtual money, you buy it and run it on a real account waiting for the day-by-day profit. Usually, it comes, like the backtest presumes, but sometimes it is not!
Why? Why an algorithm that made you money for a long time becomes unprofitable? Can an algorithm that made only trades with a maximum 2% drawdown for more than ten years blow your account? Yes, it is possible! The year 2022, when I write this article, is the best year to prove that algorithms working irreproachable for more than ten years are not good enough. There are so many today, and they are still for sale online for big money. To find the answer to how this is possible, you can read a lot nowadays. Some believers in the conspiracy theory will say that the brokerage companies are acting against the traders. Others suggest it is about the big central banks collaborating and making unpredictable decisions to move the money from small accounts into their huge and unlimited accounts. Others will also sustain that it is about some market makers organized into superior forums deciding who to win or not. Well, these are only stories for children, and none of the above are true in our case. Not at all!
An algorithm is a finite set of rules, a priori defined, made to solve a specific task. In our case, a trading algorithm is made to transform the market input data into trading decisions such as buy, sell or stay away from the market risk. The trading algorithm receives the historical quote price data, apply different computation and transformation functions, and builds the specific trading signals. Every trading algorithm has its own parameter set. These are coefficients set and optimized using the previous market data to obtain the maximum profitability and minimum values for the drawdown into a past time interval. The software authors optimize their algorithms using two, three, five, or more years of historical market data to obtain a better algorithm. In this way, a trading algorithm includes the market behavior through its parameters. In other words, it is made for a specific time interval and particular market behavior to perform well.
What in this world can assure you that tomorrow's market behavior will be the same as today or yesterday? Nothing!
Which mathematical principles can sustain that tomorrow the market will behave like in the last ten years? No one!
What facts from real life can guarantee you that tomorrow's events will be the same as in the last fifty years? None!
We think the market is stable. We want to be. This is a hypothesis. We hope the current market conditions are the same for at least a short time. But this hypothesis is not valid from time to time. There is no mathematical guarantee that the price action will be the same, as there is no motive to trust that humans, nature, or hazards will act the same way every day. The market price movement depends mainly on human behavior. It depends on all global decisions, economic facts, and natural or geopolitical events. It also mainly depends on all the market participants, their beliefs, ideas, fear, or trust. All of these variables are unbounded through the hazard theory. Consequently, the price action has a significant degree of unpredictability and is also unbounded.
An algorithm optimized for the last period will perform well if the market works like in the past, but it will not perform if the market changes significantly. The algorithm will deliver the same results only if the market price action is within the same limits and behavior as in the quotes series used for the procedure optimization. When an unprecedented event is happening, like a pandemic, war, economic crisis, or other significant events that can dramatically change the investment appetite, the algorithms can record substantial losses instead of a profit. The market change is responsible for this fact.
new files added 2026-02-24 12:47:37 -05:00			`### Automated trading algorithms`

			Using expert advisors is a simple task today. The fantastic development of MetaTrader 4 and 5 allows us to spend no more than a couple of minutes to download, test, and run an automatic software to trade our money. Thousands of expert advisors made by hundreds of authors are available online, and to buy one, you need no more than two clicks. A positive backtest report makes you feel that you met the right software. Then, you install it on a demo account and run it for a while. If it is working with virtual money, you buy it and run it on a real account waiting for the day-by-day profit. Usually, it comes, like the backtest presumes, but sometimes it is not!
			Why? Why an algorithm that made you money for a long time becomes unprofitable? Can an algorithm that made only trades with a maximum 2% drawdown for more than ten years blow your account? Yes, it is possible! The year 2022, when I write this article, is the best year to prove that algorithms working irreproachable for more than ten years are not good enough. There are so many today, and they are still for sale online for big money. To find the answer to how this is possible, you can read a lot nowadays. Some believers in the conspiracy theory will say that the brokerage companies are acting against the traders. Others suggest it is about the big central banks collaborating and making unpredictable decisions to move the money from small accounts into their huge and unlimited accounts. Others will also sustain that it is about some market makers organized into superior forums deciding who to win or not. Well, these are only stories for children, and none of the above are true in our case. Not at all!
			An algorithm is a finite set of rules, a priori defined, made to solve a specific task. In our case, a trading algorithm is made to transform the market input data into trading decisions such as buy, sell or stay away from the market risk. The trading algorithm receives the historical quote price data, apply different computation and transformation functions, and builds the specific trading signals. Every trading algorithm has its own parameter set. These are coefficients set and optimized using the previous market data to obtain the maximum profitability and minimum values for the drawdown into a past time interval. The software authors optimize their algorithms using two, three, five, or more years of historical market data to obtain a better algorithm. In this way, a trading algorithm includes the market behavior through its parameters. In other words, it is made for a specific time interval and particular market behavior to perform well.
			`What in this world can assure you that tomorrow's market behavior will be the same as today or yesterday? Nothing!`
			`Which mathematical principles can sustain that tomorrow the market will behave like in the last ten years? No one!`
			`What facts from real life can guarantee you that tomorrow's events will be the same as in the last fifty years? None!`
			We think the market is stable. We want to be. This is a hypothesis. We hope the current market conditions are the same for at least a short time. But this hypothesis is not valid from time to time. There is no mathematical guarantee that the price action will be the same, as there is no motive to trust that humans, nature, or hazards will act the same way every day. The market price movement depends mainly on human behavior. It depends on all global decisions, economic facts, and natural or geopolitical events. It also mainly depends on all the market participants, their beliefs, ideas, fear, or trust. All of these variables are unbounded through the hazard theory. Consequently, the price action has a significant degree of unpredictability and is also unbounded.
			An algorithm optimized for the last period will perform well if the market works like in the past, but it will not perform if the market changes significantly. The algorithm will deliver the same results only if the market price action is within the same limits and behavior as in the quotes series used for the procedure optimization. When an unprecedented event is happening, like a pandemic, war, economic crisis, or other significant events that can dramatically change the investment appetite, the algorithms can record substantial losses instead of a profit. The market change is responsible for this fact.