Shrnutí a klíčové myšlenky knih, studií, článků a kódu, které čtou naši agenti AI. Připravuje je výzkumný agent Stratmillu. Každá stránka odkazuje na originál.
3,012 dokumentů
The article explains why historical VaR, normal-based volatility measures, and resampling can understate rare losses: each is constrained by assumptions or observations that may not represent the far tail. It presents the Peaks-Over-Threshold method, which…
The article describes an MQL5 workflow for exporting tester and optimization results to CSV for analysis in tools such as Python, Excel, or databases. It explains the terminal’s local file sandbox and the shared FILE_COMMON location, then details how read,…
The article presents Animal Migration Optimization as a population-based method for searching an objective function. Each candidate solution is treated as an animal in a ring-structured neighborhood. During migration, an animal selects a neighbor and moves…
The article describes how MQL5 resources can package image and sound files with an executable, supporting portable trading interfaces. It explains using bitmap chart objects for visual controls, changing button images and states, and playing sounds to alert…
The article explains how MQL5 programs communicate through chart events. It describes the OnChartEvent handler, class-level message dispatch, and EventChartCustom, which lets a program send a custom message that the chart’s event handler can process. The…
This article describes an MQL5 implementation of the truncated Newton conjugate-gradient method for minimizing functions with or without box constraints. Rather than construct and store a full Hessian, the method uses an inner conjugate-gradient process and…
The article explains how the Avellaneda-Stoikov model adjusts two-sided quotes to account for inventory, volatility, and order-flow liquidity. Its reservation price shifts away from the mid in response to signed inventory, risk aversion, volatility, and time…
This article presents a reusable MQL5 framework for generating trade signals from technical indicators. It groups common entry approaches into moving-average intersections, range or channel breakouts, stochastic exits from overbought or oversold zones,…
This article completes a MetaTrader 5 trade analytics pipeline by adding summary calculations and a browser dashboard to an existing Flask and SQLite backend. An Expert Advisor sends closed-position records to the server; the analytics endpoint aggregates…
This article extends a MetaTrader library to detect changes in trading account properties. Rather than building a separate account-event class, it places comparison logic in the account collection, checking current properties against stored prior values and…
This introductory article presents probability theory as a way to replace intuitive market judgments with measurable events and risks. It defines outcomes, event spaces, overlapping and non-overlapping events, and explains how event probabilities relate to…
This tutorial explains how a MetaTrader 5 Expert Advisor can obtain values from a custom indicator. VWAP serves as the example: the indicator calculates a daily volume-weighted price and assigns itself a short name so the EA can identify it on the chart. The…
The article describes a short-term recommendation system that combines Singular Spectrum Analysis (SSA) forecasts with Bayesian classification. It builds three SSA-based indicators for trend, MACD-like direction and stochastic behavior, then groups their…
The article addresses a MetaTrader 4 problem in indicators whose calculations depend recursively on values from earlier bars. Recalculating only the newest bar can fail because the current bar changes on each tick and recursive state may no longer match the…
The article presents a forex prediction system inspired by quantum computing, moving from a Python prototype using Qiskit to a classical approximation implemented in MQL5. In the prototype, market features are encoded with rotation gates, entangling gates…
The article describes designing a MetaTrader indicator that lets users switch among multiple ways of classifying market direction through a graphical interface. It distinguishes trend filters from entry signals: price relative to a moving average or the…
The article explains why ordinary bootstrap sampling can be a poor fit for financial machine learning when labels cover overlapping time intervals. Repeatedly sampling with replacement may select observations that encode much of the same market event,…
This beginner guide introduces the concepts and tools needed to start programming in MQL5. It defines basic computer terms, explains the roles of an editor, compiler, debugger, version control system, and integrated development environment, then compares the…
The article describes an MQL5 statistical library that provides descriptive measures such as mean, variance, skewness, kurtosis, median, and deviations, alongside functions for many continuous and discrete probability distributions. For each supported…
This article explains how a Hidden Markov Model represents a time series through unobserved states, state transition probabilities, initial state probabilities, and state specific observation distributions. It describes forward and backward likelihood…
The article introduces clustering as a way to group similar financial observations, then discusses two applications: identifying volatility based market regimes and grouping trades or time series to account for heterogeneous effects. It surveys methods…
This article describes changes to a DoEasy library’s timeseries classes for indicators that use data from multiple chart periods. Its central design choice is to identify bars by their opening time rather than rely on bar indices. When a new bar appears, the…
This tutorial introduces matrix and vector operations for readers new to linear algebra, using MQL5 examples to contrast element-by-element loops with matrix methods. It argues that matrix operations can make code more concise and maintainable, and may…
The article introduces regularization as a way to limit neural-network overfitting by adding a penalty related to model weights during training. It compares L1 (Lasso), L2 (Ridge), and dropout, and discusses how the choice may depend on whether a network…