Articles on data analysis and statistics in MQL5

In this article we present yet another custom MQL5 Signal Class that we are labelling ‘CSignalBTreeBayesian’. We are marrying the algorithm of a balanced tree with a neural network that is built on Bayesian principles to formulate yet another custom signal testable independently or with other signals thanks to the MQL5 Wizard.

We implement a symbol resolution framework that abstracts broker naming differences in MetaTrader 5. Using a persistent mapping store, layered resolution with validation, a hash-indexed registry, and a cache, it returns selectable symbols with live market data and logs unresolved cases. Practically, you can deploy the same EA across brokers and keep symbol access consistent at low runtime cost.

Although the modules already interact with each other properly, an error occurs when trying to use the mouse pointer in the replay service. We need to fix this before moving on to the next step. Additionally, we will fix an issue in the mouse indicator code. So this version will be finally stable and properly polished.

For this article we switch to a custom MQL5 Wizard class that examines entry Signals. Our custom class is ‘CSignalDSUDBN’ this time around, and is coded by combining the Disjoint Set Union algorithm with a Deep Belief network. As has been the case throughout these series, our model is testable with MQL5 Wizard-Assembled Expert Advisors that can be tuned with different trailing stops and money management classes.

A statistically significant backtest is not proof of a robust edge. This article presents a three-part validation battery in Python that consumes an MQL5 trade-level CSV export. A sign-randomization permutation test evaluates whether the Sortino reflects real directional skill, bootstrap BCa intervals assess metric stability, and Monte Carlo trade-order shuffling tests sequence dependence of drawdowns. The results feed a five-condition framework for deployment decisions.

The article presents a new approach to creating trading systems based on quantum principles and artificial intelligence. The author describes the development of a unique neural network that goes beyond classical machine learning by combining quantum mechanics with modern AI architectures.

MQL5's file system operates within a strict sandbox. Understanding its access flags and path resolution rules is the foundation of any reliable export pipeline. This article builds a CCSVExporter class that handles file creation, safe appending, and error recovery. It also covers CSV parsing, field tokenization, concurrent access conflicts, and write-buffering strategies for high-frequency optimization runs.

In this article, we shift from Python research to native MQL5 engineering. We build the first module of the MMAR library: a shared constants header, an SVD-based OLS regression class, a Generalized Hurst Exponent estimator, and the partition analysis engine that computes the partition function, extracts tau(q), estimates H via zero-crossing interpolation, and scores multifractality through three diagnostic tests. Tested on 500,000 bars of EURUSD M10, the engine correctly classifies the data as multifractal in under four seconds. Part 4 of an eight-part series. Part 5 fits the tau(q) curve to four candidate distributions via the Legendre transform.

Standard MQL5 risk tools read risk from recent history and miss how heavy the downside tail can be. We implement Extreme Value Theory in MetaTrader 5: a Peaks‑Over‑Threshold fit of the Generalized Pareto Distribution via ALGLIB, a live indicator that reports EVT VaR/ES and tail shape, and an EA that sizes positions from the tail estimate. A controlled backtest illustrates reduced drawdown for unchanged entries.

This article extends single-candlestick analysis to ordered double-candlestick patterns using an MQL5 script. The script encodes candles into symbols, extracts every consecutive two-symbol sequence (treating Aa and aA as different), counts occurrences and percentages, and writes sorted frequency tables to a text file. Readers can quickly identify the most recurrent transitions by symbol, timeframe, and lookback for further statistical testing.

Batch normalization is the pre-processing of data before it is fed into a machine learning algorithm, like a neural network. This is always done while being mindful of the type of Activation to be used by the algorithm. We therefore explore the different approaches that one can take in reaping the benefits of this, with the help of a wizard assembled Expert Advisor.

In this article, we will finally solve the problems with the simulation of ticks on a one-minute bar so that they can coexist with real ticks. This will help us avoid problems in the future. The material presented here is for educational purposes only. Under no circumstances should the application be viewed for any purpose other than to learn and master the concepts presented.

We are beginning to implement the connection between Excel and MetaTrader 5, but first we need to understand some key points. This way, you won't have to rack your brains trying to figure out why something works or doesn't. And before you frown at the prospect of integrating Python and Excel, let's see how we can (to some extent) control MetaTrader 5 through Excel using xlwings. What we demonstrate here will primarily focus on educational objectives. However, don't think that we can only do what will be covered here.

SEAL (Self-Evolving Adaptive Learning) is a system for the continuous adaptation of large language models (LLMs) for algorithmic trading, designed to address the problem of rapid model degradation in changing markets. Instead of periodic retraining, which takes hours and erases old patterns, SEAL learns from every closed trade, maintains priority memory for important examples, and automatically initiates incremental fine-tuning when accuracy drops or a market regime change occurs.

The article establishes a reproducible method to measure portfolio risk for multiple symbols using MQL5 matrices and OpenBLAS. It covers computing log returns, building a covariance matrix, and evaluating wᵀΣw instead of summing individual variances. A complete script prints naive versus true volatility and the cross‑term contribution, enabling you to detect when correlated instruments inflate exposure beyond single‑asset estimates.

Downloading international monetary fund data in Python: Mining IMF data for use in macroeconomic currency strategies. How can macroeconomics help an ordinary and an algorithmic trader?

For this article we look at a custom MQL5 Wizard class for Trailing Stops. Our implemented custom class ‘CTrailingConvexHullGRU’, is built from merging the Convex Hull algorithm with a GRU network. As always we seek to develop a model that is testable with MQL5 Wizard-Assembled Expert Advisors and can be tuned with various Money Management and entry Signals classes. Our testing is with the 'Envelopes' and the RSI classes for Signal.

The article introduces the Mapper pipeline in MQL5 by implementing the two fundamental components: CTDAMapperFilter (lens) and CTDAMapperCover (overlapping intervals). It explains three lens options—eccentricity, density, and coordinate—plus cover parameters (resolution and gain), and demonstrates how a price point cloud is reduced to one value per point and interval memberships. Readers obtain ready inputs for subsequent clustering and graph construction.

The article extends MicroStructure_Foundation.mqh with a MicrostructureAnalysis struct and five functions that decompose M1 price variation into a quoted spread proxy, Roll-implied spread, OHLC-based noise ratio, order imbalance, and an adverse selection component. A wrapper populates these fields and links them to the volatility suite from Part 4. Empirical thresholds come from 602 NQ E-mini NY sessions (Jan 2024–Jun 2026), helping you gate volatility signals, size risk, and recognize spread-driven frictions.

We integrate eleven one-minute microstructure measurements from Parts 2–6 into a composite regime label with confidence and direction. A rule-based RegimeClassifier() assigns one of six regimes—Normal, Stressed, Noisy, Informed, Trending, Mean-Reverting—using empirically derived thresholds from 514 NQ M1 sessions (May 2024–May 2026). The deliverable includes MARKET_REGIME, RegimeAnalysis, and PopulateRegimeAnalysis(), enabling position sizing, stop placement, and signal filtering from a single call.

After a break in development and improvement of the service used for replay/simulator, we are resuming work on it. Now that we've abandoned the use of resources like terminal globals, we'll have to completely restructure some parts of it. Don't worry, this process will be explained in detail so that everyone can follow the development of our service.

We build a CSV exporter for MQL5 custom indicators that preserves the exact values seen on the chart. The script creates the indicator handle with iCustom, waits for BarsCalculated, aligns buffers to CopyRates, and writes a locale-safe CSV that pandas loads with parsed dates and NaN for warm-up bars. It addresses compile-time argument limits, jagged-array workarounds, and EMPTY_VALUE handling, enabling reliable Python backtests without re-coding the indicator.

The article delivers MQL5 implementations of FIGARCH and HARCH and updates the volatility library for long‑memory processes. It provides code for Hurst and GPH testing, parameter setup (truncation and horizons), and scripts for fitting, forecasting, and simulations. Readers learn how to apply and compare the models on market data to select an appropriate specification.

A GPH‑based estimator for d, the key ARFIMA parameter, is added to MicroStructure_Foundation.mqh. GPHEstimator() computes d via log‑periodogram regression, while PopulateARFIMAAnalysis() stores d with an R² confidence score and validates the theoretical relationship H = d + 0.5. An empirical study on 72 US100 M1 sessions confirms pooled d = −0.006, consistent with the random walk boundary established in Part 2.

The article presents a practical foundation for shape analysis of price series in MQL5. It implements Takens time‑delay embedding to build a phase‑space point cloud and computes the full pairwise distance matrix under selectable norms. The CTDAPointCloud and CTDADistance classes are provided with a demo script that embeds chart data and outputs results, preparing inputs for downstream topological tools.

Aggregate metrics alone do not reveal how a trade sequence manages risk. This MQL5 tool analyzes closed positions to flag four structural patterns: post-loss volume escalation, overlapping same-direction entries, asymmetric payoffs, and a classical risk-of-ruin figure. The results are merged into a configurable A-F grade with concise recommendations to guide further review.

With the multifractal parameters from Part 2 in hand, this article builds the full MMAR process. We construct the multiplicative cascade for trading time, generate Fractional Brownian Motion via Davies-Harte FFT, and combine both into X(t) = B_H[theta(t)]. A 1000-path Monte Carlo simulation produces the volatility forecast, which we then pit against GARCH on the same EURUSD M5 data. Does Mandelbrot's fractal architecture outforecast Engle's conditional variance framework? Part 3 of a eight-part series leading to a native MQL5 library and Expert Advisor.

The article examines slice sampling — an adaptive MCMC algorithm that automatically adjusts its sampling parameters. Its effectiveness is demonstrated using Bayesian linear and logistic regression models, and the results are compared with classical frequentist methods.