Articles on MetaTrader 5 integration using MQL5

A population-based optimization algorithm inspired by a controversial and little-studied phenomenon - the mechanism of human dreams. Agent groups with different "memory", cosine-wave modulation of motion, and an unusual 99/1 phase distribution — learn how these features affect the optimization efficiency of your trading strategies.

The article describes implementation of the A3 algorithm - a metaheuristic optimization method inspired by chemical processes - in MQL5. Only two adjustable parameters, compactness and a small population, ensure high operating speed with sufficient quality of solutions.

This article introduces a frequency-analysis framework for encoded candlestick patterns in MQL5. By transforming candlesticks into alphabetic symbols, historical price action can be analyzed as a statistical sequence rather than a visual chart. Using GBPUSD and Gold across multiple timeframes, the study examines the occurrence frequency of individual candlestick types, identifies dominant market structures, and reveals the symmetry between bullish and bearish price movements. The results establish a quantitative foundation for pattern discovery and prepare the way for analyzing multi-candlestick sequences and their predictive potential in algorithmic trading systems.

In the previous article, I showed how to proceed in order to add a query mechanism. This was needed so that, inside MQL5 code, you could fully use SQL and retrieve results using an SQL SELECT query. But there is still one last function we need to implement. This is the DatabaseReadBind function. Since understanding it properly requires a slightly more detailed explanation, it was decided to cover it not in the previous article, but in today's article. So, since the topic will be fairly extensive, let us proceed directly to the next section.

MetaTrader 5 is well suited for AI trading because it combines market data, MQL5 development, Python research, ONNX models, Strategy Tester, VPS, and the MQL5.community ecosystem into a single workflow. This article demonstrates a practical path from AI prompts to structured signals, working with code via the AI Assistant in MetaEditor, a quality model, a custom-created Expert Advisor, testing, and a controllable launch of a trading system.

The article presents the Competitive Learning Algorithm (CLA), a new metaheuristic optimization method based on simulating the educational process. The algorithm organizes the population of solutions into classes with students and teachers, where agents learn through three mechanisms: following the best in the class, using personal experience, and sharing knowledge between classes.

The article explores the possibility of using regression models in algorithmic trading. Regression models, unlike binary classification, allow for the creation of more flexible trading strategies by quantifying predicted price changes.

The article presents a detailed analysis of the Exchange Market Algorithm (EMA) inspired by the behavior of stock market traders. The algorithm simulates stock trading, where market participants with varying levels of success employ different strategies to maximize profits.

In today's article we will begin studying the use of SQL in MQL5 code. We will also look at how to create a database. Or, more precisely, how to create a SQLite database file using the features built into MQL5. We will also see how to create a table, and then how to establish a relationship between tables by using primary and foreign keys. All of this, once again, will be done with MQL5. We will see how easy it is to create code that can later be migrated to other SQL implementations by using a class that helps hide the implementation being created. And, most importantly, we will see that at various points we may face the risk that something will go wrong when using SQL. This happens because, in MQL5 code, SQL code will always be placed inside a string.

In the previous article, we completed the necessary introduction to SQL. And, in my opinion, we properly clarified what we wanted to show and explain about SQL. This was done so that anyone who comes to look at the market replay/simulation system being built can at least get an idea of what may be happening there. The point is that there is no sense in programming things that SQL handles perfectly.

What if an optimization algorithm could remember its past journeys and use that memory to find better solutions? BSA does just that – balancing exploration with revisiting the tried and true. In this article, we reveal the secrets of the algorithm. A simple idea, minimum parameters and a stable result.

This article develops an object-oriented framework for MQL5 indicators by evolving a primitive example into reusable modules. It formalizes partial buffer recalculation in OnCalculate, moves logic into header-based classes (CAppliedPrice, CSma), and introduces CSubIndiBase, CIndicatorBase, and a registry to centralize requirements. You get portable components, isolated inputs, and clean buffers with minimal boilerplate, making new indicators faster to assemble and easier to maintain.

In this article, we will touch upon the intriguing topic of fractal analysis and market forecasting using machine learning. These are just the first steps towards exploring the diverse fractal structures that form on financial price charts. We will use the correlation to find patterns and the CatBoost algorithm to classify these patterns.

The article explores one of the most interesting non-gradient optimization algorithms, which learns to understand the geometry of the objective function. We will focus on the classical implementation of CMA-ES with a slight modification - replacing the normal distribution with the power one. We will thoroughly examine the math behind the algorithm, as well as practical implementation, and check where CMA-ES is unbeatable and where it should be avoided.

In this article, we replace the embedded CSV snapshot with a SQLite layer that persists calendar events and triggered trade IDs across restarts. The database lives in the common terminal folder and is shared by live charts and the strategy tester, so both modes read the same data without recompiling. An on-demand downloader with a canvas progress bar fetches history from the calendar API and stores it for offline reuse.