Articles on machine learning in trading
Creating AI-based trading robots: native integration with Python, matrices and vectors, math and statistics libraries and much more.
Find out how to use machine learning in trading. Neurons, perceptrons, convolutional and recurrent networks, predictive models — start with the basics and work your way up to developing your own AI. You will learn how to train and apply neural networks for algorithmic trading in financial markets.
Add a new article
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Defining your Edge (Part 1): Using a Discrete Fourier Transform and a Spiking Neural Network in a Trading Robot
In this article we make the case for pairing the Discrete Fourier Transform with a Spiking Neural Network in a Trading Robot. The Fourier Transform helps represent data as oscillations instead of its raw values. To govern how we interpret these cycles, we engage a Spiking Neural Network that unlike regular networks, uses time dependent electrical charges to accumulate potential and only "spike" when a target threshold is met. Combining these two engines allows us better control on the timing of discrete market movements, that in theory should give us entry signals with rigorous mathematical confirmation.
Overcoming The Limitation of Machine Learning (Part 1): Lack of Interoperable Metrics
There is a powerful and pervasive force quietly corrupting the collective efforts of our community to build reliable trading strategies that employ AI in any shape or form. This article establishes that part of the problems we face, are rooted in blind adherence to "best practices". By furnishing the reader with simple real-world market-based evidence, we will reason to the reader why we must refrain from such conduct, and rather adopt domain-bound best practices if our community should stand any chance of recovering the latent potential of AI.
Population optimization algorithms: Resistance to getting stuck in local extrema (Part II)
We continue our experiment that aims to examine the behavior of population optimization algorithms in the context of their ability to efficiently escape local minima when population diversity is low and reach global maxima. Research results are provided.
MQL5 Wizard Techniques you should know (Part 55): SAC with Prioritized Experience Replay
Replay buffers in Reinforcement Learning are particularly important with off-policy algorithms like DQN or SAC. This then puts the spotlight on the sampling process of this memory-buffer. While default options with SAC, for instance, use random selection from this buffer, Prioritized Experience Replay buffers fine tune this by sampling from the buffer based on a TD-score. We review the importance of Reinforcement Learning, and, as always, examine just this hypothesis (not the cross-validation) in a wizard assembled Expert Advisor.
Resampling techniques for prediction and classification assessment in MQL5
In this article, we will explore and implement, methods for assessing model quality that utilize a single dataset as both training and validation sets.
Overcoming The Limitation of Machine Learning (Part 3): A Fresh Perspective on Irreducible Error
This article takes a fresh perspective on a hidden, geometric source of error that quietly shapes every prediction your models make. By rethinking how we measure and apply machine learning forecasts in trading, we reveal how this overlooked perspective can unlock sharper decisions, stronger returns, and a more intelligent way to work with models we thought we already understood.
Adaptive Social Behavior Optimization (ASBO): Schwefel, Box-Muller Method
This article provides a fascinating insight into the world of social behavior in living organisms and its influence on the creation of a new mathematical model - ASBO (Adaptive Social Behavior Optimization). We will examine how the principles of leadership, neighborhood, and cooperation observed in living societies inspire the development of innovative optimization algorithms.
Neural Networks in Trading: Piecewise Linear Representation of Time Series
This article is somewhat different from my earlier publications. In this article, we will talk about an alternative representation of time series. Piecewise linear representation of time series is a method of approximating a time series using linear functions over small intervals.
Forecasting exchange rates using classic machine learning methods: Logit and Probit models
In the article, an attempt is made to build a trading EA for predicting exchange rate quotes. The algorithm is based on classical classification models - logistic and probit regression. The likelihood ratio criterion is used as a filter for trading signals.
Integrating MQL5 with data processing packages (Part 3): Enhanced Data Visualization
In this article, we will perform Enhanced Data Visualization by going beyond basic charts by incorporating features like interactivity, layered data, and dynamic elements, enabling traders to explore trends, patterns, and correlations more effectively.
Overcoming The Limitation of Machine Learning (Part 9): Correlation-Based Feature Learning in Self-Supervised Finance
Self-supervised learning is a powerful paradigm of statistical learning that searches for supervisory signals generated from the observations themselves. This approach reframes challenging unsupervised learning problems into more familiar supervised ones. This technology has overlooked applications for our objective as a community of algorithmic traders. Our discussion, therefore, aims to give the reader an approachable bridge into the open research area of self-supervised learning and offers practical applications that provide robust and reliable statistical models of financial markets without overfitting to small datasets.
Neural Networks in Trading: Memory Augmented Context-Aware Learning for Cryptocurrency Markets (Final Part)
The MacroHFT framework for high-frequency cryptocurrency trading uses context-aware reinforcement learning and memory to adapt to dynamic market conditions. At the end of this article, we will test the implemented approaches on real historical data to assess their effectiveness.
Neural networks made easy (Part 61): Optimism issue in offline reinforcement learning
During the offline learning, we optimize the Agent's policy based on the training sample data. The resulting strategy gives the Agent confidence in its actions. However, such optimism is not always justified and can cause increased risks during the model operation. Today we will look at one of the methods to reduce these risks.
Low-Frequency Quantitative Strategies in MetaTrader 5 (Part 5): Pre-Backtest Evaluation of Machine-Learning-Generated Signals Through Formulaic Alphas
The article shows how to evaluate machine-learning alphas before a full backtest by expressing them as formulaic alphas. We compute Information Coefficient (IC), Rank IC, Information Ratio (ICIR), and t-statistics to quantify forecasting strength and stability. A MetaTrader 5 backtest illustrates differences versus execution-dependent tests, and a Python parser facilitates reproducible calculations and bulk screening.
Biological neuron for forecasting financial time series
We will build a biologically correct system of neurons for time series forecasting. The introduction of a plasma-like environment into the neural network architecture creates a kind of "collective intelligence," where each neuron influences the system's operation not only through direct connections, but also through long-range electromagnetic interactions. Let's see how the neural brain modeling system will perform in the market.
Animal Migration Optimization (AMO) algorithm
The article is devoted to the AMO algorithm, which models the seasonal migration of animals in search of optimal conditions for life and reproduction. The main features of AMO include the use of topological neighborhood and a probabilistic update mechanism, which makes it easy to implement and flexible for various optimization tasks.
Artificial Bee Hive Algorithm (ABHA): Theory and methods
In this article, we will consider the Artificial Bee Hive Algorithm (ABHA) developed in 2009. The algorithm is aimed at solving continuous optimization problems. We will look at how ABHA draws inspiration from the behavior of a bee colony, where each bee has a unique role that helps them find resources more efficiently.
Anarchic Society Optimization (ASO) algorithm
In this article, we will get acquainted with the Anarchic Society Optimization (ASO) algorithm and discuss how an algorithm based on the irrational and adventurous behavior of participants in an anarchic society (an anomalous system of social interaction free from centralized power and various kinds of hierarchies) is able to explore the solution space and avoid the traps of local optimum. The article presents a unified ASO structure applicable to both continuous and discrete problems.
Analyzing binary code of prices on the exchange (Part II): Converting to BIP39 and writing GPT model
Continuing tries to decipher price movements... What about linguistic analysis of the "market dictionary" that we get by converting the binary price code to BIP39? In this article, we will delve into an innovative approach to exchange data analysis and consider how modern natural language processing techniques can be applied to the market language.
Population optimization algorithms: Spiral Dynamics Optimization (SDO) algorithm
The article presents an optimization algorithm based on the patterns of constructing spiral trajectories in nature, such as mollusk shells - the spiral dynamics optimization (SDO) algorithm. I have thoroughly revised and modified the algorithm proposed by the authors. The article will consider the necessity of these changes.
Ensemble methods to enhance numerical predictions in MQL5
In this article, we present the implementation of several ensemble learning methods in MQL5 and examine their effectiveness across different scenarios.
Successful Restaurateur Algorithm (SRA)
Successful Restaurateur Algorithm (SRA) is an innovative optimization method inspired by restaurant business management principles. Unlike traditional approaches, SRA does not discard weak solutions, but improves them by combining with elements of successful ones. The algorithm shows competitive results and offers a fresh perspective on balancing exploration and exploitation in optimization problems.
Causal inference in time series classification problems
In this article, we will look at the theory of causal inference using machine learning, as well as the custom approach implementation in Python. Causal inference and causal thinking have their roots in philosophy and psychology and play an important role in our understanding of reality.
The case for using a Composite Data Set this Q4 in weighing SPDR XLY's next performance
We consider XLY, SPDR’s consumer discretionary spending ETF and see if with tools in MetaTrader’s IDE we can sift through an array of data sets in selecting what could work with a forecasting model with a forward outlook of not more than a year.
Reimagining Classic Strategies (Part IV): SP500 and US Treasury Notes
In this series of articles, we analyze classical trading strategies using modern algorithms to determine whether we can improve the strategy using AI. In today's article, we revisit a classical approach for trading the SP500 using the relationship it has with US Treasury Notes.
MQL5 Wizard Techniques you should know (Part 34): Price-Embedding with an Unconventional RBM
Restricted Boltzmann Machines are a form of neural network that was developed in the mid 1980s at a time when compute resources were prohibitively expensive. At its onset, it relied on Gibbs Sampling and Contrastive Divergence in order to reduce dimensionality or capture the hidden probabilities/properties over input training data sets. We examine how Backpropagation can perform similarly when the RBM ‘embeds’ prices for a forecasting Multi-Layer-Perceptron.
Neural Networks in Trading: Disentangling Structured Components (SCNN)
We invite you to explore the innovative SCNN framework, which takes time series analysis to a new level by clearly separating data into long-term, seasonal, short-term, and residual components. This approach significantly improves forecasting accuracy by allowing the model to adapt to complex and changing market dynamics.
Neural Networks in Trading: Exploring the Local Structure of Data
Effective identification and preservation of the local structure of market data in noisy conditions is a critical task in trading. The use of the Self-Attention mechanism has shown promising results in processing such data; however, the classical approach does not account for the local characteristics of the underlying structure. In this article, I introduce an algorithm capable of incorporating these structural dependencies.
MetaTrader 5 Machine Learning Blueprint (Part 17): CPCV Backtesting — From Python Model to Tick-Level Evidence
We bridge Python-native artifacts to MQL5 for tick-accurate CPCV backtesting. The export script converts the ONNX model, calibrator, feature spec, and path masks to flat files, while the expert advisor rebuilds features, performs ONNX inference with calibration, and trades on real ticks. The Strategy Tester runs each combinatorial path, and Python aggregates per-path equities into a path Sharpe distribution to assess robustness after spread, slippage, and commission.
The Group Method of Data Handling: Implementing the Combinatorial Algorithm in MQL5
In this article we continue our exploration of the Group Method of Data Handling family of algorithms, with the implementation of the Combinatorial Algorithm along with its refined incarnation, the Combinatorial Selective Algorithm in MQL5.
Neural Networks in Trading: Injection of Global Information into Independent Channels (InjectTST)
Most modern multimodal time series forecasting methods use the independent channels approach. This ignores the natural dependence of different channels of the same time series. Smart use of two approaches (independent and mixed channels) is the key to improving the performance of the models.
Neural Networks in Trading: Point Cloud Analysis (PointNet)
Direct point cloud analysis avoids unnecessary data growth and improves the performance of models in classification and segmentation tasks. Such approaches demonstrate high performance and robustness to perturbations in the original data.
Building Volatility models in MQL5 (Part I): The Initial Implementation
In this article, we present an MQL5 library for modeling volatility, designed to function similarly to Python's arch package. The library currently supports the specification of common conditional mean (HAR, AR, Constant Mean, Zero Mean) and conditional volatility (Constant Variance, ARCH, GARCH) models.
Dolphin Echolocation Algorithm (DEA)
In this article, we take a closer look at the DEA algorithm, a metaheuristic optimization method inspired by dolphins' unique ability to find prey using echolocation. From mathematical foundations to practical implementation in MQL5, from analysis to comparison with classical algorithms, we will examine in detail why this relatively new method deserves a place in the arsenal of researchers facing optimization problems.
Neural Networks in Trading: Generalizing Time Series Without Data-Specific Dependence (Mamba4Cast)
In this article, we introduce the Mamba4Cast framework and take a closer look at one of its key components: timestamp-based positional encoding. The article shows shows how time embedding is formed taking into account the calendar structure of the data.
Spurious Regressions in Python
Spurious regressions occur when two time series exhibit a high degree of correlation purely by chance, leading to misleading results in regression analysis. In such cases, even though variables may appear to be related, the correlation is coincidental and the model may be unreliable.
Category Theory in MQL5 (Part 19): Naturality Square Induction
We continue our look at natural transformations by considering naturality square induction. Slight restraints on multicurrency implementation for experts assembled with the MQL5 wizard mean we are showcasing our data classification abilities with a script. Principle applications considered are price change classification and thus its forecasting.
MQL5 Wizard Techniques you should know (Part 32): Regularization
Regularization is a form of penalizing the loss function in proportion to the discrete weighting applied throughout the various layers of a neural network. We look at the significance, for some of the various regularization forms, this can have in test runs with a wizard assembled Expert Advisor.
Neural Networks in Trading: Hierarchical Dual-Tower Transformer (Final Part)
We continue to build the Hidformer hierarchical dual-tower transformer model designed for analyzing and forecasting complex multivariate time series. In this article, we will bring the work we started earlier to its logical conclusion — we will test the model on real historical data.
Category Theory in MQL5 (Part 21): Natural Transformations with LDA
This article, the 21st in our series, continues with a look at Natural Transformations and how they can be implemented using linear discriminant analysis. We present applications of this in a signal class format, like in the previous article.