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.
5,018 documents
The article introduces Mamba4Cast, a time-series forecasting framework that combines state-space sequence modeling with Prior-data Fitted Networks. Instead of pretraining on one instrument’s historical prices and then fine-tuning, the approach learns from…
The article examines how to reduce false and delayed signals from a conventional moving average crossover strategy. Its baseline uses fast and slow averages for direction, with ATR-based stop-loss and take-profit levels. A manual refinement adds a…
The article presents an Expert Advisor signal that turns RSI, Stochastic, and price context into seven directional scores. Each score is bounded around a neutral midpoint, then isotonic regression maps its ranking to an empirical probability. An optional…
The article explains Decision Transformer as an offline reinforcement learning method that frames action selection as autoregressive sequence modeling. Its trajectory representation interleaves return-to-go, state, and prior action tokens; at inference, a…
This article explains two ways to move market data between MetaTrader 5 and Python. One uses a TCP socket server in Python and an MQL5 client: the terminal sends recent closing prices, Python fits a linear regression against time, and the client receives the…
The article presents an MQL5 interpretation of FinAgent, a tool-augmented framework that combines market analysis, reflection, memory, auxiliary tools, and decision support. For time-series analysis, the implementation uses a transformer model; classical…
The article presents the Bonobo Optimizer as a population-based method for numerical optimization. Candidate solutions are bonobos, and the current best candidate is the alpha. The search combines three offspring-generation approaches: movement toward the…
The article surveys adaptive neural network optimizers, explaining how AdaGrad, RMSProp, Adadelta, and Adam adjust parameter updates using accumulated gradient information. It describes AdaGrad’s shrinking learning rate, RMSProp’s exponentially weighted…
This article introduces design choices in population optimization, focusing on binary genetic algorithms. It compares real-valued and binary encodings of optimization parameters, describing the flexibility of real numbers and the bitwise operations enabled…
The article explores using “fan” templates to turn recent candlestick closing prices into inputs for perceptrons and neural networks. Each input is a rounded point distance between the close of the oldest candle in a window and closes at selected positions…
The article describes a workflow that connects Python research with MetaTrader 5 data and trading tools. It covers setting up an isolated Python environment, installing data analysis and machine learning libraries, connecting to the terminal, and retrieving…
The article treats preprocessing as a tunable part of a machine-learning trading pipeline and compares transformations in a feedback-controlled strategy. Its baseline uses twelve input features without scaling; after an initial 90-day period, the controller…
The article explains Extralonger, a neural network architecture designed to combine spatial relationships and temporal patterns in long-horizon forecasting. It motivates the method through an analogy between traffic networks and financial markets, where…
This technical guide introduces MQL5’s vector and matrix types as tools for representing ordered numerical data and expressing linear algebra operations. It explains how to initialize vectors with preset values or a filling function, resize and copy them,…
The article explains four ways to measure information in the sequence of trade directions within a bar: Shannon entropy for symbol frequencies, plug-in entropy for repeated blocks, Lempel-Ziv complexity for compressibility, and Kontoyiannis entropy for…
The article investigates whether Visa’s monthly Spending Momentum Index series can help classify future EUR/USD direction. It combines headline, discretionary, and non-discretionary spending measures with monthly price data, examines plots and correlations,…
The document explains a genetic algorithm that represents each candidate solution as a chromosome of real-valued genes. It describes a population split between parent and offspring groups, fitness-based sorting, selection, crossover, mutation, and…
This guide walks through diagnosing MQL5 compilation errors and using the Strategy Tester to evaluate and optimize an Expert Advisor. Its debugging examples cover common issues such as mismatched types, missing punctuation or braces, incorrectly declared…
The article presents a MQL5 implementation of the Attraos approach to time-series forecasting, combining phase-space reconstruction, neural-network components, and adaptive state-space modeling. It explains how lagged observations are represented through the…
The article introduces AdaBoost as an ensemble classification method that combines weak learners, commonly decision stumps or shallow trees. It explains that training examples receive changing weights: misclassified cases gain emphasis in later rounds, while…
This concluding article describes an implementation of a Spatial-Sequential Convolutional Neural Network for financial time-series forecasting in MQL5. Its central subject is the encoder, which separates long-term trend, seasonal, short-term, and spatial…
The article describes an MQL5 trading system organized around memory units, individual agents called brains, and a collective decision layer. Each memory unit associates an encoded market state with an action and tracks reward, estimated value, activations,…
The article introduces functors as mappings between categories that preserve objects, relationships, composition, and identity. It applies this framing to daily tide observations from Monterey, California, and NASDAQ volatility data, proposing that one…
The article explains how fuzzy logic can soften rigid trading rules by assigning indicator values degrees of membership in linguistic groups such as buy, neutral, and sell. It contrasts this approach with crisp logic and sketches the stages of fuzzy…