Articles on trading system automation in MQL5

Generative Adversarial Networks are a pairing of Neural Networks that train off of each other for more accurate results. We adopt the conditional type of these networks as we look to possible application in forecasting Financial time series within an Expert Signal Class.

Explore whether financial markets are truly random by recreating Larry Williams’ market behavior experiments using MQL5. This article demonstrates how simple price-action tests can reveal statistical market biases using a custom Expert Advisor.

In the last piece, we concluded our look at the pairing of the gator oscillator and the accumulation/distribution oscillator when used in their typical setting of the raw signals they generate. These two indicators are complimentary as trend and volume indicators, respectively. We now follow up that piece, by examining the effect that supervised learning can have on enhancing some of the feature patterns we had reviewed. Our supervised learning approach is a CNN that engages with kernel regression and dot product similarity to size its kernels and channels. As always, we do this in a custom signal class file that works with the MQL5 wizard to assemble an Expert Advisor.

As the model is trained based on the experience reproduction buffer, the current Actor policy moves further and further away from the stored examples, which reduces the efficiency of training the model as a whole. In this article, we will look at the algorithm of improving the efficiency of using samples in reinforcement learning algorithms.

We continue to integrate methods proposed by the authors of the Attraos framework into trading models. Let me remind you that this framework uses concepts of chaos theory to solve time series forecasting problems, interpreting them as projections of multidimensional chaotic dynamic systems.

We continue to explore the analysis and forecasting of time series in the frequency domain. In this article, we will get acquainted with a new method to forecast data in the frequency domain, which can be added to many of the algorithms we have studied previously.

This article applies the Optimal Trading Rule from AFML Chapter 13 to set profit targets and stop-losses without in-sample calibration. We model post-entry P&L with a discrete Ornstein–Uhlenbeck process, run a 100,000-path search, and implement Python, multiprocessing, and a Numba @njit parallel kernel (242× faster). The result is an optimal (PT, SL) under three forecast specifications, constrained by the prop-firm daily loss limit.

In this article, we will look at changes that will allow the replay/simulation system to operate more efficiently and securely. I will also not leave without attention those who want to get the most out of using classes. In addition, we will consider a specific problem in MQL5 that reduces code performance when working with classes, and explain how to solve it.

We revamp our earlier articles on testing trade setups with the MQL5 Wizard by putting a bit more emphasis on input data quality, cleaning, and handling. In the earlier articles we had looked at a lot of custom signal classes, usable by the wizard, so we now shift our focus to a custom trailing class, given that exiting is also a very important part in any trading system. Our broad theme for this particular piece data-efficiency and the O(1) range-query; the core ‘tech’ is MQL5, SQLite, Python-Polars; the Algorithm is the Sparse-Table while we will seek validation from the ATR Indicator.

Let's outline the main stages of the EA development. One of the first things to be done will be to optimize a single instance of the developed trading strategy. Let's try to collect all the necessary information about the tester passes during the optimization in one place.

We continue to study algorithms for extracting features from a point cloud. In this article, we will get acquainted with the mechanisms for increasing the efficiency of the PointNet method.

When people started creating the first systems capable of computing, everything required the participation of engineers, who had to know the project very well. We are talking about the dawn of computer technology, a time when there were not even terminals for programming. As it developed and more people got interested in being able to create something, new ideas and ways of programming emerged which replaced the previous-style changing of connector positions. This is when the first terminals appeared.

This article shows how to automate Larry Williams Smash Day reversal patterns in MQL5 within a structured context. We implement an Expert Advisor that validates setups over a limited window, aligns entries with Supertrend-based trend direction and day-of-week filters, and supports entry on level cross or bar close. The code enforces one position at a time and risk-based or fixed sizing. Step-by-step development, backtesting procedure, and reproducible settings are provided.

In this article, we complete the first phase of developing our replay and simulation system. Dear reader, with this achievement I confirm that the system has reached an advanced level, paving the way for the introduction of new functionality. The goal is to enrich the system even further, turning it into a powerful tool for research and development of market analysis.

We invite you to get acquainted with the DADA framework, which is an innovative method for detecting anomalies in time series. It helps distinguish random fluctuations from suspicious deviations. Unlike traditional methods, DADA is flexible and adapts to different data. Instead of a fixed compression level, it uses several options and chooses the most appropriate one for each case.

One major challenge is managing multiple chart windows of the same pair running the same program with different features. Let's discuss how to consolidate several integrations into one main program. Additionally, we will share insights on configuring the program to print to a journal and commenting on the successful signal broadcast on the chart interface. Find more information in this article as we progress the article series.

We continue studying the Hierarchical Vector Transformer method. In this article, we will complete the construction of the model. We will also train and test it on real historical data.

We continue our last article on DDPG with MA and stochastic indicators by examining other key Reinforcement Learning classes crucial for implementing DDPG. Though we are mostly coding in python, the final product, of a trained network will be exported to as an ONNX to MQL5 where we integrate it as a resource in a wizard assembled Expert Advisor.

The article introduces the dialectical algorithm (DA), a new global optimization method inspired by the philosophical concept of dialectics. The algorithm exploits a unique division of the population into speculative and practical thinkers. Testing shows impressive performance of up to 98% on low-dimensional problems and overall efficiency of 57.95%. The article explains these metrics and presents a detailed description of the algorithm and the results of experiments on different types of functions.

The article presents the Big Bang - Big Crunch method, which has two key phases: cyclic generation of random points and their compression to the optimal solution. This approach combines exploration and refinement, allowing us to gradually find better solutions and open up new optimization opportunities.

Loss Function is the key metric of machine learning algorithms that provides feedback to the training process by quantifying how well a given set of parameters are performing when compared to their intended target. We explore the various formats of this function in an MQL5 custom wizard class.

In the previous last article within this series, we looked at the Atom-Motif Contrastive Transformer (AMCT) framework, which uses contrastive learning to discover key patterns at all levels, from basic elements to complex structures. In this article, we continue implementing AMCT approaches using MQL5.

The article outlines a practical data pipeline for quantitative analysis based on Parquet storage, Hive-style partitions, and DuckDB. It details migrating selected SQLite tables to Parquet, structuring market data by source, symbol, timeframe, and date, and querying it with SQL window functions. A Golden Cross example illustrates cross‑symbol evaluation of forward returns. Accompanying Python scripts handle data download, conversion, and execution.

Reinforcement Learning is one of the three main tenets in machine learning, alongside supervised learning and unsupervised learning. It is therefore concerned with optimal control, or learning the best long-term policy that will best suit the objective function. It is with this back-drop, that we explore its possible role in informing the learning-process to an MLP of a wizard assembled Expert Advisor.

The article outlines a practical data pipeline for quantitative analysis based on Parquet storage, Hive-style partitions, and DuckDB. It details migrating selected SQLite tables to Parquet, structuring market data by source, symbol, timeframe, and date, and querying it with SQL window functions. A Golden Cross example illustrates cross‑symbol evaluation of forward returns. Accompanying Python scripts handle data download, conversion, and execution.

The article considers a metaheuristic Artificial Ecosystem-based Optimization (AEO) algorithm, which simulates interactions between ecosystem components by creating an initial population of solutions and applying adaptive update strategies, and describes in detail the stages of AEO operation, including the consumption and decomposition phases, as well as different agent behavior strategies. The article introduces the features and advantages of this algorithm.

The Fractal Adaptive Moving Average (FrAMA) and the Force Index Oscillator are another pair of indicators that could be used in conjunction within an MQL5 Expert Advisor. These two indicators complement each other a little bit because FrAMA is a trend following indicator while the Force Index is a volume based oscillator. As always, we use the MQL5 wizard to rapidly explore any potential these two may have.

In this discussion, we will carefully extend the existing Dialog library to incorporate theme management logic. Furthermore, we will integrate methods for theme switching into the CDialog, CEdit, and CButton classes utilized in our Admin Panel project. Continue reading for more insightful perspectives.

This article proposes an asset screening process for a statistical arbitrage trading strategy through cointegrated stocks. The system starts with the regular filtering by economic factors, like asset sector and industry, and finishes with a list of criteria for a scoring system. For each statistical test used in the screening, a respective Python class was developed: Pearson correlation, Engle-Granger cointegration, Johansen cointegration, and ADF/KPSS stationarity. These Python classes are provided along with a personal note from the author about the use of AI assistants for software development.

A backtest shows only one path among many possible outcomes. This MQL5 script performs 1000 bootstrap Monte Carlo resamples of a trade P&L series, draws a percentile fan chart on the chart via CCanvas, and reports probability of ruin, value at risk, and 95th‑percentile worst drawdown. The result is a practical view of path risk and drawdown exposure beyond a single equity curve.

The Deep-Q-Network is a reinforcement learning algorithm that engages neural networks in projecting the next Q-value and ideal action during the training process of a machine learning module. We have already considered an alternative reinforcement learning algorithm, Q-Learning. This article therefore presents another example of how an MLP trained with reinforcement learning, can be used within a custom signal class.

We follow up on our last article, where we introduced the indicator pair of the Ichimoku and the ADX, by looking at how this duo could be improved with Supervised Learning. Ichimoku and ADX are a support/resistance plus trend complimentary pairing. Our supervised learning approach uses a neural network that engages the Deep Spectral Mixture Kernel to fine tune the forecasts of this indicator pairing. As per usual, this is done in a custom signal class file that works with the MQL5 wizard to assemble an Expert Advisor.