Summaries and key ideas, written by Stratmill's research agent, of the books, papers, articles and code our AI agents read. Each page links to its original.
4,574 documents
The article describes an MQL5 Expert Advisor for a session range breakout strategy. It defines the range using the high and low from midnight to 6 AM, then looks for a breakout during a later time window. A Break of Structure confirmation, based on price…
The document presents a way to make trading forecasts more interpretable by translating fitted machine-learning models into algebraic expressions. Its example uses technical features derived from price bars, including returns, volatility, RSI, MACD, and…
This article describes incorporating an adapter-tuned GPT-2 model into a MetaTrader Expert Advisor for trading decisions. It compares three deployment approaches: converting the model to ONNX for in-platform inference, launching Python scripts from the EA,…
The article presents a MetaTrader 5 dashboard that combines structural direction from several timeframes into a normalized Harmony Index ranging from bearish to bullish. Each enabled timeframe receives a bias based on a three-bar staircase rule, with…
This installment extends a MetaTrader 5 library that stores historical timeseries for multiple symbols and periods. Its central problem is keeping non-chart symbols current without repeatedly refreshing every series on a timer, which wastes work when no…
The article describes a Python library that mirrors several MQL5 Standard Library trade classes for use with the MetaTrader 5 Python package. It outlines wrappers for account, symbol, order, history, position, deal, and terminal information, along with a…
The article outlines a Python and MetaTrader 5 workflow for searching historical forex bars for recurring sequences of price directions. It converts close-to-close changes into up or down labels, tests sequences of varied lengths, and records whether the…
This article describes a file-based filter that reads a Forex Factory calendar CSV into structured event records and checks whether the current broker server time falls within a configured news window for a symbol. Its components parse dates, times,…
This article explains a classic hedge cycle that alternates buy and sell positions after each stop loss. Each new position doubles the previous lot size, aiming for a fixed net gain when price eventually reaches one of two take profit levels. The author also…
The article compares account-level netting and hedging in MetaTrader 5. Under netting, deals on a symbol aggregate into one position: same-direction trades increase volume, while opposite trades reduce, close, or reverse it. Under hedging, trades create…
This interview profiles a four-person team managing a shared forex signal account. The traders describe combining individual technical analyses, checking them against economic news and the expected daily range, then having two members select the pair and…
The article describes an MQL5 Expert Advisor that combines Q-learning, cross-pair correlation adjustments, and market features drawn from three timeframes. It represents market observations as an indexed state and scores a set of actions that includes…
The article examines whether swap payments can contribute to returns and describes locking positions across accounts with different swap schedules. It argues that holding offsetting positions can reduce exposure to price direction while the combined swaps…
The article proposes a simple chart exercise for practicing short-term market-direction forecasts. At the start of a candle, the program displays up and down arrows; the user removes one to record a prediction. When the candle closes, the program compares…
This article explains that a trend depends on both the chart timeframe and the period being examined, then compares several ways to classify direction. These include price relative to a moving average, ordered moving-average fans, successive ZigZag highs and…
The article treats price movement as velocity: price change divided by elapsed time. It compares averages calculated over whole candles with measurements on smaller timeframes and tick data. A candle average is easy to interpret but hides the path inside the…
This article explains how a Deep Q Network can evaluate three trading patterns built from TRIX, a smoothed momentum indicator, and Williams Percent Range (WPR), an overbought and oversold measure. It outlines the value based reinforcement learning setup:…
The article presents a method for selecting Forex positions with positive rollover income while accounting for exchange-rate returns and volatility. It describes scoring eligible currency pairs using normalized estimates of market return, swap, and…
This interview recounts one participant's approach to automated trading and his robot in the 2012 Automated Trading Championship. The featured system was a simple moving-average strategy, generated with the MQL5 Wizard and adjusted for competition rules. Its…
The article proposes organizing economic calendar observations with database tables and category-theory mappings. Its motivation is that calendar events differ in units, formats, geography, and meaning: some publish numeric indicators, while others are…
The article describes a two-stage workflow for measuring costs and using them to set more realistic triple-barrier labeling thresholds. An MQL5 script gathers broker-specific spread history, hourly spread averages, swap settings, and symbol details into a…
This guide combines DeMark Sequential with Murray-Gann price levels to interpret trend development and possible reversals. Sequential begins with a directional setup based on closes relative to earlier bars, confirms a run of qualifying bars, waits for an…
This article explains an attempt to adapt N-BEATS, a residual time-series forecasting architecture, for a forex Expert Advisor in MetaTrader 5. The model decomposes a series through stacked blocks that successively explain historical input and contribute…
The article proposes using cross-validation predictions to identify training examples that a classifier repeatedly mislabels. A primary CatBoost model is trained on examples judged more predictable, while a second classifier learns across the wider dataset…